Biolitika

DevOpsDays 2025

Roman Luštrik — Sat, 13 Sep 2025 07:13:00 +0200

Lintering using tox ignores Python version

Roman Luštrik — Fri, 10 Nov 2023 09:52:00 +0100

When trying to run linters using tox, it would keep use Python 3.11 to run. But this isn't what's defined in my project configuration that says it should be Python versions between and including 3.8 and 3.10 nor the version used in my virtual environment.

➜ tox -e linters
...
Processing ./.tox/.tmp/....tar.gz
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Installing backend dependencies ... done
  Preparing metadata (pyproject.toml) ... done

ERROR: Package 'mypackage' requires a different Python: 3.11.6 not in '<3.11,>=3.8'

The specific configuration (tox.ini, setup.cfg and pyproject.toml) worked for others and testing server, but not for me, so I started poking around my system. The first clue was when I checked tox --version which said 4.11, BUT, it was from folder .../python3.11/site-packages. When I tried to remove the global version using pip uninstall tox, pip complained that this isn't possible because the package might have been installed using the system's package manager. Sure enough, after sudo dnf remove tox and installation of tox in the local environment, lintering now goes through.

Zakaj se ne prijavim na razpise v javni upravi

Roman Luštrik — Thu, 22 Dec 2022 09:44:00 +0100

Gospa Alenka Kolar,

na internetsih sem zasledil vaš komentar, kjer pozivate razvijalce, naj se prijavimo na delovna mesta, da bomo sprememba, ki jo želimo videti. Da ne bomo jamrali, ko se bomo naslednjič borili v zdravstvu s pomanjkanjem podatkov, nerodnimi rešitvami ali poznim izvidom.

Čisto kot zanimivost občasno sledim objavam prostih delovnih mest v javni upravi, sam pa sem bil že večkrat povabljen, da se prijavim na to ali ono delovno mesto. Razpisodajalci mi razlagajo kakšne borbe so imeli znotraj kolektiva, da so za nekega informatika izborili neko nepredstavljivo visok plačni razred. Če bi rekli, da se ljudje gnetejo pred vhodom porabnikov javnega denarja, bi resnico še kar nategovali. Menim, da lahko kot nekdo, ki gledam od zunaj v javno izložbo, ponudim nekaj uvidov zakaj bi imela javna uprava probleme s pridobivanjem kvalificiranega kadra.

Naj začnem s tem, da seveda najverjetneje ne predstavljam (povprečnega) IT-jevca, da so vsi pogledi samo moji in kot bomo vsak pri sebi ugotovili, domala čudni. Po izobrazbi sem biolog in biostatistik, priučen programer, po prepričanju pa znanstvenik. Zlahta si je predstavljati, da na svet gledam in sodim skozi merokratska očala. Za vse slovnične in pravopisne napake se že vnaprej opravičujem, vendar pa mislim, da je iz spodaj napisanega razvidno kaj je avtor želel povedati.

Na prvo žogo je seveda lahko reči, da je problem služb v javnem sektorju višina plače. Oziroma pomanjkanje višine. Plače so z vidika ponudbe v industriji res mizerne, saj je ponujen bruto v javni upravi velikokrat celo nižji od neta, ki ga nam ponujajo delodajalci v zasebnem sektorju. S tako plačo si človek danes ne more več privoščiti povprečnega stanovanja in tistih dveh tednov v Umagu. Pa vendar gremo težko mimo tega, kar imenujemo "delovni pogoji", ki so sestavljeni tudi iz plače, ampak obstaja še cela mavrica drugih pogojev, ki bi jih lahko razvrstili po pomembnosti.

Zame in za mnoge moje znance predstavljata upogljiv delavnik in delo od doma izmed bolj pomembnih pogojev, ki si jih na delovnem mestu želimo. Tega v razpisnih pogojih v neindustriji navadno ne zasledimo. Tudi po pričanjih kolegov, ki delajo v javni upravi je to nekaj, za kar se morajo praktično boriti in biti skoraj izobčeni v kolektivu. Zaradi narave dela mi je nedojemljivo zakaj česa takega delodajalec ne bi omogočil. Sicer pa so razpisi slabo pripravljeni. Opisi nalog so skoraj neobstoječi. Iz vsega (ne)napisanega ne dobim nobene motivacije, da bi delal v kolektivu, ker sploh ne vem kaj bom delal.

Da pa zaključim še z najpomembnejšim razlogom zakaj se ne prijavim na delovna mesta v javni upravi. Menim, da problem našega zdravstva in javne uprave ni zaradi pomanjkanja kadra v IT-ju, ampak so to zgolj posledice, ki jih vidimo na daleč.

Problemi so večji in predvsem organizacijske in politične narave. Že samo Univerzitetni klinični center v Ljubljani ima preko trideset takih in drugačnih sistemov za obvladovanje podatkov. Kako pride do takega stanja? Verjetno ne zaradi IT-jevcev ampak sistema, ki jih je prisilil, da so razvili ali nabavili različne rešitve. Stvari, ki bi morale trajati minute, če ne sekunde, se v takih sistemih lahko razvlečejo na tedne. To ni dobro delovno okolje.

Ko ustanova objavi razpis za programsko rešitev je razpis očitno prirejen ponudniku, ki s tem ali onim jé kekse na jahti ali igra briškulo na Mavriciju. Znotraj ustanov poteka neznosna politična bitka, zaposleni so navzkriž skregani, vse to pa dela ljudi apatične. To ni dobro delovno okolje.

Zdravniki jamrajo o slabih programskih rešitvah. Porabili smo na desettisoče evrov in skoraj desetletje življenja mladega človeka, ga prevelik del svojega časa klika po Google Maps™. Zdravnik naj dela s pacientom za njegov dolgotrajni dobrobit, ne pa bojevanje z birokracijo in birokrati. V takih delovnih pogojih zdravnike razumem, da ne želijo delati.

To sklepanje lahko prenesemo tudi na vodilne položaje upravljavcev zdravstvenih in drugih ustanov. V zdravstvenih ustanovah so to pogosto zdravniki, ki smo jih leta izobraževali, z delom so v času postali strokovnjaki na svojem področju, nato pa dvignejo sidra in postanejo managerji. Menim, da so za upravljavska mesta bolj primerni za to usposobljeni ljudje. Upravljanje s premoženjem inštituta, priprava razpisov in druge birokratske dejavnosti verjetno niso del učnega načrta na Fakulteti za medicino. Se pa morda motim. Vse to na pol amatersko delo pa ima za posledico niža kvaliteto storitev in povečuje trenje v samem sistemu, kot smo jim priča. To niso dobri obeti, ki jih vidim v potencialnem delodajalcu.

Primerov (še ne) dokazane korupcije sploh ne bom začenjal naštevati.

Gospa Kolar, problem našega zdravstva ni IT, problem našega zdravstva je sistemske in politične narave. Če, bo to rešeno z vrha, ne od znotraj. Do takrat pa bo še slabše preden bo postalo bolje.

How to tile windows in thirds (wide screens)

Roman Luštrik — Mon, 12 Sep 2022 10:30:00 +0200

Window management was made easy for me using gTile. A handy accelerator ("key binding") is to tile windows in thirds. This comes especially handy in wider screens where I can tile three windows next to each other. To do this I used Preset resize

3x1 1:1 1:1, 2:1 2:1, 3:1 3:1

If you map this to an accelerator, you will be able to cycle tile window to first, second or third third of the window.

Problem when waking Fedora from sleep on a ThinkPad Carbon X1

Roman Luštrik — Thu, 14 Jul 2022 11:03:00 +0200

If there's anything I really hate is booting up my system every day. Sorting windows to appropriate screens and virtual desktops is a hassle. Which is why I really love the sleep feature. The problem was, when I tried to wake the computer up, my secondary screen would not work anymore. Connection is via USB-C, the computer detects it, it's charging over it, but there is no picture. Even in settings, no secondary display was detected. This was solved by running

inxi -Fzx

and the picture appeared instantly. Don't ask me how this works, it's just something I tried when I ran diagnostics from this page.

Notes on intermittent fasting

Roman Luštrik — Sun, 12 Jun 2022 11:11:00 +0200

The why

After listening to the Effects of fasting & time restricted eating on fat loss & health podcast episode, twice, I realized that I need to distill my thoughts and take-home messages into a more accessible form. These notes may not make sense to you, so I encourage you to listen to the whole episode and see if they make sense then.

When we partake in diets, this can be due to many different reasons. Most people think of weight loss, but you can also be dieting for fat loss, muscle mass, organ health, genome health (not sure what specifically was meant here), epigenome (I guess methylation of the genome?), to improve exercising, mood, lifespan, inflammation and cognition. I'll touch on a few of these and expand with notes I made.

What is intermittent fasting?

Generally, this is considered a period when you are (not) eating. You can think of it as time of fasting, or, time of eating. This is why I may refer to it as time restricted feeding in this document. This also goes in line with the circadian rhythm of your metabolism and gene activity. Huberman reports that about 80 % of genes have a diurnal cycle. Technically, we're crepuscular animals and addressing this with also your (not) eating window would kind of makes sense.

When you sleep, body is "cleaning metabolic debris" (including autophagy), which can be helped by intermittent fasting.

When we eat, digestion happens (resulting in shit happens), which is an active process. During the process, among other things, inflammatory biomarkers go up and when going into fasting state, these biomarkers subside. This inflammatory process is important when up-taking and digesting food. I can imagine if this happens around the clock that can be very stressful on the body and mind, also contributing to bad mood.

The dos and don'ts

You should have the feeding window set to the more active part of the day. It's very important that you keep it consistent, as by shifting it you will move your body into another time zone. It can take two to three days to get your body (and mind?) back in line. For many people, 7-9 hours of feeding time seems in line with their social life as well as accommodate the acceptable duration between meals. For example, if you start eating at noon, your last bite should be no later than eight in the evening. Many people who do 8 hour eating period actually do more like 9. If you are aiming for 8 hours, do 7 instead. Ideally, you would not eat one hour after waking up, preferably more, and at least two to three hours before sleep.

It takes 5-6 hours to go into fasting state (that is when blood glucose is low). You can accelerate the lowering of blood glucose by some light exercise (i.e. a walk), intensive exercise or using agents. Be warned that using chemical agents like berberine and metformin can have side effects like headaches.

Simple sugars will raise your blood glucose levels the most. This is followed by protein and fat having the slowest effect on raising of the glucose. This is important for breaking fast as 1 g of sugar may break your fast and a small fat heavy meal may not. This depends on the time of day and state of your body. After a marathon, a meal may not break your fast.

Effects on health and cognition

When fasting, we can detect a decrease in p-S6 and mTOR, latter being associated with aging.

It is reported to have a positive effect on gut microbiome, possibly helping IBS (irritable bowl syndrome) patients and patients with various form of colitis. It is said that it promotes the replacement of possibly offending Lactobacillus with species of bacteria associated with healthier gut, i.e. Acidobacter.

Perhaps patients with non-alcoholic fatty liver disease could reap some benefits from intermittent fasting as it promotes the creation of brown fat, which is associated to be positively correlated with the absence of NAFL. Another way of producing brown fat is through cold exposure.

In elite cyclists, time restricted feeding helped increase testosterone and cortisol levels. FWIW, low cortisol levels are suppose to be associated with depression.

Every other day fasting can produce caloric reduction but can be a challenge for at least some to adhere to the regiment.

Time restricted feeding makes it so that you don't have to think when to (not) eat, freeing you from potentially "expensive" decisions.

In states of fasting, fat is a preferred way of increasing blood glucose levels. This happens through increased activity of hepatic lipase and decreased CIDEc (which is lipolysis inhibitor).

Some people find that consuming a pinch of salt (or some lemon juice) helps them huddle through the fasting period. This possibly happens because salt can increase the blood volume.

Time restricted feeding can also help lower blood pressure.

Pimping up the new install of Fedora

Roman Luštrik — Sat, 07 May 2022 00:00:00 +0200

After a fresh install of a Fedora 35, I noticed that videos and gifs were a bit slow. After a bit of searching I found this post which gives instructions on how to install the ffmpeg system package.

sudo dnf install https://mirrors.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install ffmpeg-libs

Another things I also noticed is that Rstudio would no longer work because of the

rstudio: error while loading shared libraries: libssl.so.10: cannot open shared object file: No such file or directory

I suspect this was because I installed the RHEL7 (CentOS) version of the package. The error went away when I installed the RHEL8 version from here.

Fixing Fedora the old fashion way

Roman Luštrik — Mon, 02 May 2022 13:18:00 +0200

My regular maintenance of my Fedora instance includes doing incessant upgrading using

sudo dnf upgrade

which updates any possible packages and the kernel. This has worked fine until recently, when two of my computers started freezing during the upgrading process. Usually, I would wait about 10 minutes before hard shutting down the computer and powering back up without any obvious detremental effects.

But a few days ago, I cycled the computer after a freeze after about a minute. When trying to powering back up, I was welcomed by the emergency mode saying

Cannot open access to console, the root account is locked.
See sulogin(8) man page for more details.

Press Enter to continue.

Needless to say, anything I tried to do at this stage would not bear any fruit, resulting in a lot of FAILED messages, from failure to mount /boot/efi to Failed to start Load Kernel Module.

The problem was resolved by booting into a live USB Fedora and installing a fresh install. The only thing to watch for is to not reformat the /home folder. After the system was up and running, I was welcommd by having all my setting being retained, i.e. all Firefox windows were as I left them before the crash. When I installed Slack, all channels were loaded without going through the tedious logging in using two-factor authentication. PyCharm also kept all the sweet settings I set.

The key to fixing the computer is just pouring copious quantities of your favorite wine. And having a backup computer also helps.

Remember processed entities in Python

Roman Luštrik — Sat, 04 Sep 2021 10:38:00 +0200

There exist not-so-edge cases when you're processing data but your code fails for some reason or another. There are many solutions around this. Complexity increases fault tolerance but also the cost of setting things up. For instance, you could create a Celery task queue or a Redis database for persistent storage to process tasks, but this can be somewhat involved. It can also not be feasible if you're working in a contained environment where installing a new package or piece of software would take a Jira ticket and days of waiting. Also, if you're doing a one-off thing, anything other than simple may be an overkill.

To remember which entities have been already processed, my initial solution included writing this information into a pickle file. Before processing an entity, I would reread the pickle file and check if the entity has been processed previously. This seems very Flintstony, so I thought I'd use the collective hive mind and asked this question on a local developer community slack channel. A bunch of solutions rushed in, including Redis and Celery, but the one that got my attention the most was the shelve object used for persistent storage.

It's very easy and intuitive to work with. It's just a matter of setting up a shelf as it were and putting things on it. It's nice to tear the shelf down once you're done using it. It's even prepackaed with Python, so no additional installation of packages is needed.

Here's a small script that demonstrates the functionality. A function run_process processes all the files in folder. First, I add three files and process them. Then I add two more files and process the whole set of files again. Notice that the files that have already been processed are now skipped. This is the result of the script:

['In file1']
['In file2']
['In file3']
Skipping ./file1
Skipping ./file2
Skipping ./file3
['In file4']
['In file5']

The script in its entirety or see this gist:

"""
Off the shelf way to skip already processed entities

This scripts processes files, remembers which files have been already
processed and skips them on the next run.
"""
import glob
import os
import shelve


def create_files(file_list):
    """Creates specified files on disk with boring contents."""
    for obj in file_list:
        with open(obj, mode="wt") as to_disk:
            to_disk.write(f"In {os.path.basename(obj)}")
    return None


def process_file(x):
    """Reads file and prints its contents."""
    with open(file=x, mode="rt") as in_file:
        content = in_file.readlines()
        print(content)


def list_files(path):
    """List all files in a specified path."""
    files = glob.glob(os.path.join(path, "file*"))

    return sorted(files)


# noinspection PyBroadException
def run_process(files):
    """
    Iterate through specified files and process them if not already
    done in the previous run.
    """
    for obj in files:
        try:
            key = os.path.basename(obj)
            if key not in shelf:
                process_file(x=obj)
                shelf[key] = "OK"
            else:
                print(f"Skipping {obj}")
        except Exception as e:
            print(f"Failed ({e})")

    return f"Processed {len(files)} files."


# This is a file where memories of processed files go.
CACHE_FILE = "processed.cache"

shelf = shelve.open(filename=CACHE_FILE)

create_files(file_list=["file1", "file2", "file3"])

available_files = list_files(path=".")

run_process(files=available_files)

create_files(file_list=["file4", "file5"])

available_files = list_files(path=".")

run_process(files=available_files)

# Close the shelf
shelf.close()

# Remove any files that may have been created during the execution of
# this script.
for file in available_files:
    os.unlink(file)

os.unlink(CACHE_FILE)

Working with Notion API from R

Roman Luštrik — Mon, 21 Jun 2021 19:30:00 +0200

When searching for a solution where I could store some flat files as a database, Notion came up. The nice thing about it is that it offers an API to most of its functionality. At the time of this writing this is still in beta, but hopefully it will become even more powerful in the future.

The functionality I'm concerned with right now is interacting with the databases. The boilerplate code to start communicating via the Notion API is

library(httr)
library(jsonlite)

NOTION_KEY = "enter-your-key"

ask <- GET(
  url = "https://api.notion.com/v1/databases/",
  add_headers("Authorization" = paste("Bearer", NOTION_KEY),
              "Notion-Version" = "2021-05-13"))
stop_for_status(ask)
fromJSON(rawToChar(ask$content))

You should get an R list of elements returned by this databases endpoint.

To write to a database (add a row), you need to prepare an object that corresponds to the corresponding table schema. A JSON object would look like something like this:

properties: {
      Name: {
        title: [
          {
            text: {
              content: 'Tuscan Kale',
            },
          },
        ],
      },
      Description: {
        text: [
          {
            text: {
              content: 'A dark green leafy vegetable',
            },
          },
        ],
      },
      'Food group': {
        select: {
          name: '🥦 Vegetable',
        },
      },
      Price: {
        number: 2.5,
      },
    },

Using a similar logic, I wrote my call in R

pb <- list(
  parent = list(database_id = NOTION_DATABASE_ID),
  properties = list(
    Name = list(
      title = list(text = list(content = "4"))
    ),
    Genus = list(
      rich_text = list(text = list(content = "Neki"))
    ),
    Species = list(
      rich_text = list(text = list(content = "noviga"))
    )
  )
)

but this produced the following error

body failed validation. Fix one: body.properties.Name.title.id should be defined, instead was `undefined`.
body.properties.Name.title.name should be defined, instead was `undefined`.
body.properties.Name.title.start should be defined, instead was `undefined`.

Long story short, it turns out that every curly braces and array need to be a list(). This worked (notice the extra list() calls compared to the original):

pb <- list(
  parent = list(database_id = NOTION_DATABASE_ID),
  properties = list(
    Name = list(
      title = list(
        list(text = list(content = "4"))
      )
    ),
    Genus = list(
      rich_text = list(
        list(text = list(content = "Neki"))
      )
    ),
    Species = list(
      rich_text = list(
        list(text = list(content = "noviga"))
      )
    )
  )
)

send.row <- POST(
  url = "https://api.notion.com/v1/pages",
  add_headers("Authorization" = paste("Bearer", NOTION_KEY),
              "Notion-Version" = "2021-05-13"),
  body = pb,
  encode = "json"
)

Problem compiling RSQLite from source

Roman Luštrik — Sat, 22 May 2021 08:53:00 +0200

When I tried to compile RSQLite from source on Fedora 34, I was awaited by this error message.

Error installing package 'RSQLite':
===================================
* installing to library ‘/.../renv/staging/1’
* installing *source* package ‘RSQLite’ ...
** package ‘RSQLite’ successfully unpacked and MD5 sums checked
** using staged installation
** libs
g++ -m64 -std=gnu++11 -I"/usr/include/R" -DNDEBUG -I. -Ivendor
-DRSQLITE_USE_BUNDLED_SQLITE -DSQLITE_ENABLE_RTREE
-DSQLITE_ENABLE_FTS3 -DSQLITE_ENABLE_FTS3_PARENTHESIS -DSQLITE_ENABLE_FTS5
-DSQLITE_ENABLE_JSON1 -DSQLITE_ENABLE_STAT4 -DSQLITE_SOUNDEX
-DRCPP_DEFAULT_INCLUDE_CALL=false -DRCPP_USING_UTF8_ERROR_STRING
-DBOOST_NO_AUTO_PTR -DSQLITE_MAX_LENGTH=2147483647 -DHAVE_USLEEP=1
-I'/../R/x86_64-redhat-linux-gnu-library/4.0/plogr/include'
-I'/.../renv/staging/1/Rcpp/include' -I/usr/local/include  
-fvisibility=hidden -fpic  -O2 -flto=auto -ffat-lto-objects -fexceptions
-g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,
-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -fstack-protector-strong
-specs=/usr/lib/rpm/redhat/redhat-annobin-cc1  -m64  -mtune=generic
-fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection  
-c DbColumn.cpp -o DbColumn.o
In file included from pch.h:1,
                 from DbColumn.cpp:1:
RSQLite.h:9:10: fatal error: Rcpp.h: No such file or directory
    9 | #include <Rcpp.h>
      |          ^~~~~~~~
compilation terminated.
make: *** [/usr/lib64/R/etc/Makeconf:181: DbColumn.o] Error 1
ERROR: compilation failed for package ‘RSQLite’

To make it work, I think reinstalling Rcpp package made the difference. I have been upgrading my Fedora lately (bumping from F33 to F34), which leads me to believe that there may have been a change in compiler versions. What I would imagine is that something didn't play well together.

Water well outflow rate

Roman Luštrik — Sat, 15 May 2021 08:50:00 +0200

The house I live in has been built around 1939 and is accompanied by a water well that is about 6 m deep. This was dug out by hand, 0.5 m a day, added scaffolding, added concrete and so on for a few weeks. I think, but see the below figure where on the right side you can see the jagged ledges that would explain this hypothesis.

In the past (think 1990), this well has been the only source of water for the house, but ever since we've been connected to a public water source, it's been there mostly for show.

Soon we'll be rebuilding the house and rainfall from the roof will have to go somewhere. One of the options is to use this well to siphon the water into. It would be a great feature if the water would flow out of the well so as not to overflow.

In the first week of this May, there was a large rainfall and water level was about 220 cm above the top edge of the well (about 1 m under the surface). This was the ideal opportunity for me to measure how fast the water will go down. Using a tape measure, I measured the distance from the top of the well to the water level with about 5 cm accuracy. It's hard to measure precisely because I could only approximately see when the measurer hit the water level. I noted date and time and wrote it down into a Google Spreadsheet table.

To calculate the rate, I created a (generalized) linear model where I would model water level height as a function of time (seconds). In R parlance, that would be glm(water_level_height ~ time). Summary of the model is

##
## Call:
## glm(formula = razdalja_cm ~ date_time, data = xy)
##
## Deviance Residuals:
##     Min       1Q   Median       3Q      Max  
## -12.257   -3.863   -1.472    4.954   13.578  
##
## Coefficients:
##                     Estimate       Std. Error t value      Pr(>|t|)    
## (Intercept) -619585.78740010   22316.13758332  -27.76 0.00000000306 ***
## date_time         0.00038248       0.00001377   27.78 0.00000000304 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 61.77171)
##
##     Null deviance: 48160.00  on 9  degrees of freedom
## Residual deviance:   494.17  on 8  degrees of freedom
## AIC: 73.382
##
## Number of Fisher Scoring iterations: 2

If you look at the date_time coefficient, it says that the water level changes 0.00038 cm per second. That makes it about 33 cm a day. The measurements were taken in the span of 6.68 days, so the total lowering of the water level was about 221 cm (calculated backwards from the model). Which is approximately what I measured, starting at 220 cm and ending at 430 cm below water well edge.

To put this into liters, given the diameter of 100 cm, that translates to outflow of 230 L of water per day.

Percentili plač za leto 2019

Roman Luštrik — Thu, 13 May 2021 18:21:00 +0200

Na Statističnem uradu RS so bili zelo prijazn in za leto 2019 (in druge) objavili podatke o porazdelitvi bruto plač. Na sliki 2 na prejšnji povezavi si lahko ogledaš porazdelitev plač na podlagi leta 2019, na tem mestu pa ponujam manjšo razširitev tega grafa.

Stolpci predstavljajo odstotek plač v danem razredu. Bruto plača na osi x je odrezana pri 7000 EUR bruto. Rdeče pike so odveč in predstavljajo iste vrednost, kot jih prikazujejo stolpci. Moj dodatek je 1) črta, ki prikazuje interpolirane (napovedane) vrednosti med dvema pikama in 2) kumulativni izračun, ki prikazuje koliko plač je nižjih od izbrane vrednosti. Za slednjo informacijo pojdi z miško čez črto ali rdečo piko.

Upgrading Fedora 31 -> 33 on a Dell

Roman Luštrik — Sat, 26 Dec 2020 16:30:00 +0100

Upgrading operating system usually isn't a pleasant experience in my book. This is especially true for Linux machines. This experience is drawn based on playing around in Debian, Ubuntu and lately, Fedora.

Against my better judgment I decided to upgrade to the latest Fedora version (currently 33). Basically, I followed instructions such as these, where you install a plugin, upgrade package and finally the OS.

# This is a one-time operation.
$ sudo dnf install dnf-plugin-system-upgrade

$ sudo dnf upgrade --refresh
$ sudo dnf system-upgrade download --releasever=32
$ sudo dnf system-upgrade reboot

When installing 31 -> 32, I did dnf update --refresh, under Software installed everything that had to be upgraded and went through with the upgrade. Upon reboot (last command), computer's firmware started to be upgraded. It would seem there were some firmware updates. It didn't take long for the error messages to start popping up and computer automatically shutting down. Great.

After turning it on, I kept getting an assistant window which tried to correct mistakes. It would seem that unable to find a bootable device means nothing is wrong.

At this point no amount of reboots helped. Having all the backups I was not worried but I did manage to forget to create a bootable Fedora LiveUSB drive. Great again.

Calling a friend, he created the latest Fedora workstation 33 image. When we tried to boot from that USB drive, the problem was that the computer wouldn't even turn on. After searching the internet for possible answers and doing all sorts of combinations with holding down the power button and connecting/disconnecting the computer from power source, all of a sudden the computer woke up and resumed upgrading Fedora (without the USB key). What. The. Fuck.

After the upgrade finished, I followed the same three steps (upgrade, system-upgrade and reboot) for version 33 and that time around, things went smoothly.

O iskanju optimalnih ukrepov

Roman Luštrik — Tue, 08 Dec 2020 17:20:00 +0100

Smo v decembru, že več kot pol leta po tem, ko je pandemija udarila tudi v tej podalpski deželici. Na vladnem vrhu še vedno ista kombinacija političnih strank (hvala bogu, da se vsaj ta spremenljivka ni spremenila) in še vedno isti pristop k ukrepom omejevanja posledic virusa SARS-CoV-2. Bolj podrobno, zapiranje javnega življenja, omejevanje prehajanja občin, policijska ura, kraja javnega denarja pri belem dnevu. Kot v eni lepi predvidljivi vojni.

Pri vsem tem pa je po mojem mnenju problem, da ti ukrepi ne jemljejo v ozir človeške narave, torej, da smo socialna bitja. Socialnost ni samo nekaj kar slišiš na izbrani fakulteti, ampak je rezultat milijone let evolucije živčnih poti. Zato se ljudje še vedno ustrašijo, ko vidijo nekaj kači podobnega in še vedno nasmehnemo, ko slišimi radostni otroški smeh.

Preden preidem na moje videnje kako do "idealnih" ukrepov pa naj povem, da sem veroizpovedi, ki trdi, da je ključna sestavina edinih in pravih palačink, jajce. Nekateri bi rekli, da je to lažnjiva dihotomija, kot to radi rečemo v strokovnem parlansu, ampak jaz to razumem kot zatiskanje oči pred realnostjo. Samo zato, ker pogledaš stran, še ne pomeni, da se nekaj ne zgodi.

Da bi kar se da hitro prišli do optimalne kombinacije ukrepov je verjetno več pristopov.

Eden od njih bi bil, da pogledamo čez mejo in vidimo kaj deluje in kaj ne. Pri tem bi bilo potrebno prej preučiti vse možne dejavnike, ki vplivajo na ljudi kar se tiče (ne)upoštevanja ukrepov in se po možnosti razlikujejo med državami. Ker kar deluje v Nemčiji ni nujno, da bo delovalo tudi v Sloveniji.

Drugi način bi bil, da se neke prostorske enote tretira kot testno in kontrolno skupino. Najprej se moramo najprej odločiti katere ukrepe sploh spreminjati in katere ne. Tu so nam navadno v pomoč literatura in izkušnje iz drugih držav. Velik izziv je tudi kako ukrepe merimo. V nekaterih vedah so lahko meritve zelo natančne, pri merjenju družbe pa prihaja do zelo netočnih mer, ki lahko tudi zelo zamujajo zato so lahko robno uporabne.

V prvi skupini izvedemo eno kombinacijo ukrepov, na primer v eni skupini je uvedena policijska ura, v drugi pa ne. Ko preteče v naprej določena količina časa se pogleda podatke in odloči katera skupina je glede na merjenje dejavnike na boljšem. Špeganje velja, a vseeno se mora poskus izvesti do konca. Ne smemo se pustiti skušnjavi, da test prekinemo, ko zaznamo nek učinek, ki nam je všečen.

Naslednji test je lahko nova kombinacija ukrepov in tako lahko iterativno (postopoma) izboljšujemo ukrepe, ki se prilagajajo na trenutno situacijo.

Sicer bi s testiranjem morali začeti že v marcu. Razumljivo je, da je bilo takrat vse novo, smo reagirali kar se da ostro in je obrodilo sadove. Vendar pa si čez poletje nismo vzeli čas za razmislek in priprave in posledice čutimo zdaj. Iterativno testiranje bi vzelo veliko časa, ena dobra posledica pa bi bila verjetno naša kolektivna večja možnost za prilagajanje na prihajajoče preizkušnje, ki nas čakajo.

Držanje glave v pesku in ignoriranje nedelovanja ukrepov bo verjetno daleč premalo zato upam, da se Vlada čim prej odloči nekaj spremeniti (kar bi bilo enako mojemu pristopu, a še enkrat počasneje in manj zanesljivo, ker ni kontrolne oz. referenčne točke).

Seznam bizark, ki mi pridejo na oko ali uho

Roman Luštrik — Sun, 15 Nov 2020 13:33:00 +0100

Tole je nekaj bizark, ki sem jih slišal ali videl/bral na lastne oči. Primeri niso mišljeni, da bom(o) na njih kazali s prstom ali jih kakorkoli zasramovali. Kazen je že inherentna v (ne)dejanju. Ti primeri so zanimivi za razumevanje kako se ljudje spopadajo, ko so postavljeni med dva mlinska kamna (možnost, da zboli za okužbo in kaznijo, ki jo lahko izda nadležni organ) in kako trenutni (ne)ukrepi (ne)delujejo. Kolikor mi je znano za enkrat država odzivov ne preučuje sistematsko. Za posledico ne vemo kakšno je dejansko stanje. Če poznaš še kak primer, se priporočam (ga objavim anonimno). Ta zapis je bolj mišljen zame kot pa nek obsežen pregled stanja na terenu.

Če se vidite v kakem od primerov, je to zgolj naključje (razen v tistem enem primeru, ko ste dejansko vi).

Tale prihaja s Primorskega konca. Oseba dostavlja eno ključno slovensko živilo. Pri tem vestno nosi masko. Pri eni stranki ga ogovorijo, da "sej pr nas je pa ja ne nuca". Čez dva dni oseba zboli in gre na testiranje. Izve, da je pozitiven. Kot dober državljan svoji stranki sporoči o svojem zdravstvenem stanju, nakar mu po telefonu odgovorijo "aja, ja, sej pr nas smo pozitivni zdej že nekaj časa".

V podjetju si sodelavke podajajo naslov frizerke, ki incognito ureja pričeske na domu.

Prebivalko doma za starejše občane testirajo na SARS-Cov-2 in jo nato pripeljejo na slikanje. V tej točki rezultati testa še niso znani. Izkaže se, da je bila pozitivna, z njo pa je bilo v stiku več kosov medicinskega osebja. Čez nekaj dni dve administratorki z oddelka pozitivni. To vse ni bila ovira, da si čez nekaj dni ekipa ne bi postregla z narezkom in sokom, da proslavijo okroglo obletnico ene od oseb z oddelka.

V eni ljubljanski lekarni so trije farmacevti. Sicer je lekarna zračna, a majhna (kot tista kajžica iz ljudske pesmi). Farmacevtu na blagajni pada maska pod nos, ostala dva jo pa po potrebi nadeneta le, ko prideta blizu pulta.

V Štajerski prestolnici se dogajajo najrazličnejše zadeve. Par, ki sta verjela v teorijo zarote glede COVID-a, sta zbolela. On je dobil neko hujšo vročino (več podrobnosti ni na voljo), ona pa se je po pozitivnem testu sprehajala na Mariborsko Piramido, kjer jo je nekdo prepoznal. Izgovarjala se je, da se je "samo šla na klopco v park usest".

Tale prihaja iz neznanega kraja. Soseda se je, po nekaj minut pogovora, "pohvalila", da se je mož šel testirati, ker se je počutil slabo. In se je izkazalo, da je bil pozitiven. Epilog ni znan.

Na Gorenjskem neznano kje je k možakarju prišla na obisk teta. Po pol ure klepetanja teta by-the-usput omeni, da se je šla testirat in je bila pozitivna. Ampak po njenih besedah ni panike, ker nima simptomov. Po tem, ko je bila promptno odslovljena čez okno je možakar v samoizolaciji.

Nekje na Celjskem v hišici, kjer potoček žubori in krokarji pojejo svoj kra-kra, družina živi. Vsi po malem zboleli (vneto grlo, vročina), a ker vsi še vohajo in okušajo se niso odpravili na test. Posledično nemoteno hodijo v službo.

Policija z marico z vožnjo čez pločnik zapre pot dvema sprehajalcema. Sprehajata se namreč z manj kot 2 m medsebojne razdalje. Sicer živita v skupnem gospodinjstvu, ampak ker nista poročena oz. ker se to ne vidi iz uradnih evidenc, prejmeta plačilni nalog.

V vrtcu so pri enem otroku odkrili okužbo z virusom. Vsi otroci so dobili odrejeno karanteno. Ker je otrok ostal brez dnevnega varstva, je dobila mati odrejeno nadomestilo za varstvo otroka. Mož hodi v službo, ker njemu to nadomestilo ne pripada. Ker gre za majhno podjetje, je večina zaposlenih v podobni situaciji (imajo nekoga doma v karanteni) in si ne morejo privoščiti, da bi delo ustavili. Sicer vzdržujejo varno razdaljo in nosijo maske. Razen za malico, ko se nagužvajo v čajnico, kjer obedujejo in klepetajo brez mask.

Na šoli so imeli potrjen primer virusa. Otroci iz razreda, kjer je bila najdena okužba, so ostali doma. Njihovi brati in sestre pa so nemoteno hodili v isto šolo. Starši so se udeležili tradicionalni vaških dogodkov, ki jih praznujejo ob različnih obletnicah brez upoštevanja varne radalje ali nošenja maske. In večini se to ni zdelo čudno.

Gospa je zbolela za virusom SARS-CoV-2. Sicer neposredno ni bila testirana, vendar pa so bili testirani drugi člani družine in bili pozitivni. Zabavna notica: odrejena jim ni bila karakntena, nihče jih ni poklical. Vse je odvisno od presoje posameznika. In ta posameznica se je odločila, da bo šla cca 1 teden po nastopu simptomov na eno od wellness storitev. Če ste sklepali kaj je ta oseba po izobrazbi, ste tisti, ki ste ugibali mikrobiologinja, uganili pravilno.

Gospa zbolela za virusom in bila tudi testirana pozitivna. Vzorno ni več hodila v službo in se držala doma. Česar pa ne moremo reči za njune otroke, ki so še dalje hodili v službo, šolo, do sosedov... In jim to ni bilo nič čudnega. Gospa iz prve povedi dela kot medicinsko osebje v eni od slovenskih bolnic.

Gospoda je dva dni bolela glava. Ker je to eden od simptomov COVID-19, je prepričal osebno zdravnico, da mu je odredila samoplačniški test. Doma imajo namreč nekoga, ki spada v rizično skupino in jim je ta informacija pomembna. Test je bil pozitiven in gospod je seveda ostal doma, stike z osebo iz rizične skupine pa so omejili. Gospa je brez simptomov, a še vedno hodi v službo in vam bo vaš nakup obračunala v eni od poslovalnic multinacionalne trgovske verige.

Mlajša družba je praznovala uspešen zaključek obdobja enega od svojih vrstnikov. Rajali so v stanovanju, ki ni bil pretirano zračeno. Okužilo se je več deset ljudi. Mediji poročajo, da zaradi tega slabega zračenja. Nihče pa seveda ni poročal, da je bil kajen isti zvitek prepovedane droge.

Ali se spopadamo s pandemijo po najboljših močeh?

Roman Luštrik — Sun, 15 Nov 2020 12:44:00 +0100

V času pisanja smo imeli v Sloveniji dva vala okužb z virusom SARS-CoV-2 (ki povzroča bolezen COVID-19). Še vedno se spomnim kako sem na začetku prvih spomladanskih ukrepov stal pred hišo in gledal vrste pred eno izmed lokalnih poslovalnic multinacionalne trgovske verige. Če so v ZDA ljudje frantično kupovali papir za odmetavanje v straniščno školjko, pa smo Slovenci za pogubo pripravljeni z moko in kvasom. Da je bila mera polna, so nas omejili še na gibanje znotraj občin, kjer imamo stalno bivališče. Ljubljana je bila turistično že prazna, cel kup ekspatov pa se je preselilo nazaj na svoje podeželjske rezidence.

V drugem valu, ki nas je presenetil kot cestarje sneg v decembru, je vlada spet zaigrala na podobno vižo kot v prvem valu. Omejevanje dejavnosti, omejevanje na statistične regije, omejevanje na občine, omejevanje na druženje, prepovedano zbiranje in protestiranje, obvezno nošenje mask, delno zapiranje tega in onega, popolno zaprtje tega in onega, omejevanja prodaje s teh in unih polic, a ne s tistih drugih. Hkrati je tudi za nekaj časa nehala testirati in izdajati karantenske odločbe.

Če smo ukrepe v prvem valu sprejeli in vzeli zelo resno, tega verjetno ne morem reči za drugi val (osebna opažanja). Ljudje so se začeli ogibati ukrepom, iskati luknje in zanašati se ne dejstvo, da verjetno ne bodo oglobljeni. Po mojem velik del krivde ni na ljudeh samih, ampak je to tesno povezano s sistemom v katerem smo (o tem več v naslednjem prispevku). Iz prvega vala so ljudje videli kaj sledi in kaj to pomeni za njihov vsakdan. Ne vem kako prizadeti so zaradi okrnjenih vsakodnevnih prostodnevnih aktivnosti, vem pa kot dejstvo (N=5), da si ljudje ne morejo privoščiti izpada dohodka, ker veliko gospodinjstev nima privarčevanih sredstev za nekaj mesecev brez ali zmanjšanih dohodkov. Enako velja za podjetja, ki so praktično kaznovana, če imajo privarčevanih preveč sredstev (in so primorana le-ta po takih in drugačnih obvodih spreminjati v druge, manj likvidne, vire). Meni izgleda, kot da se je vlada odločila, da smo državljani kot otroci, ki nismo sposobni samostojno lulati in kakati, zato nas neuspešno poskuša upravljati na majhni skali (t.i. micromanagement). To deluje v hierarhičnih združbah kot je vojska, bojim pa se, da ljudje ne delujemo na principe in dogme, ki veljajo v strogo ustrojenih ustanovah.

V dobre pol leta odkar se že relativno aktivno ukvarjamo s pandemijo pa se, kot nekdo, ki si kruh reže z analizo podatkov, sprašujem na podlagi katerih podatkov se odločevalec odloča glede tega katere ukrepe sprejeti. Za enkrat nisem dobil še nobenega občutka, da se ti ukrepi sprejemajo na podlagi (poglobljene) analize stanja, se pa seveda lahko motim, ker nisem muha na steni sej vlade. Zadnji val ukrepov mi da celo misliti, da odločevalec ne ve točno kaj počne, ali pa jih izrablja za pedlanje kakih drugih agend, za katere pa jaz ne vem. Kako lahko po tem, ko se kazalniki novih dnevnih okužb začnejo zmanjševati, država še dodatno stopi na zavoro? Saj vemo, da je škoda narejena, ker so ti kazalniki kak slab teden za dejanskim stanjem, ne? Smrti, ki bodo nastale zaradi povečanja okužb so že "zacahnane", kot to radi rečemo na Gorenjskem. Če imamo take probleme s spopadanjem s pojavi, ki imajo zamika en teden ali mesec, kako se bomo spopadli s stvarmi, ki imajo zamika trideset let (sensu podnebne spremembe)? Vprašanje je seveda retorično.

Če ponazorim kako vidim ukrepanje vlade na slikanem primeru. Predstavljajte si, da se pelje avto po avtocesti. V njem so kakopak voznik (vlada) in sopotniki. Pelje se z največjo dovoljeno hitrostjo in ob vsakem valu voznik krepko zavira, da se avto trese (hvala bogu, da imamo ABS), zavorni diski zaškripajo, na asfaltu se celo poznajo črte, pnevmatike dajo vse od sebe, sopotniki kričijo z rokami v zraku. Avtomobil se upočasni in za las se izogne trku. Počasi spet pospeši in je nazaj na nominalni hitrosti 130 km/h. Pride drugi val, ponovno zaviranje. Ker vozniku ne izgleda, da zaviranje deluje dovolj dobro, potegne še ročno zavoro. Iz zavornih čeljusti se kar kadi, mašina teče v prazno s huronskim brenčanjem. Do kam lahko na tak način pripelje vlada prebivalstvo? Črnega Kala? Postojne? Goričkega?

Aja, če mislite, da smo državljani v tej zgodbi sopotniki, se motite. Državljani smo motor, zavore, pnevmatike, pepelnik. Tisti z nekoliko več sreče radio.

Torej, koliko takih bomo potrpeli preden se na družbi poznajo zajede v čustvenem, fizičnem in gospodarskem zdravju? Jaz menim, da tale virus še ne bo izkoreninjen tako hitro kot si nekateri bolj ali manj goreči zagovorniki biotech-a predstavljajo. Do takrat pa potrebujemo bolj trajnostne ukrepe, ki jih bomo lahko izvajali na on/off s kar se da majhnimi prekinitvami. Kaj moramo postoriti, da bomo na avtocesti le spustili plin, ne pa na polno zavirali? Kateri ukrepi so to? Da izveš več, klikni sem.

Build Pelican static website using GitHub actions

Roman Luštrik — Thu, 06 Aug 2020 20:18:00 +0200

GitHub enables the delivery of static contents using GitHub Pages. That is, you create a GitHub repository and serve it as a regular static website. For most use cases I need, this is more than enough.

I use Pelican to create a static website from .md files. This content resides within the master branch of my repository and consists of theme folder, content (where my blog posts are saved), Makefile and some configuration files. Since I'm serving my website to a custom domain (and not just romunov.github.io), I also have a ./content/extra/CNAME entry with biolitika.si for the re-routing to work.

To get the GitHub workflow running, I created a .github/workflows/build-blog.yml file with the following content:

name: Deploy biolitika blog
on:
  push:
    branches:
      - master
jobs:
  build_job:
    name: Deploy blog
    runs-on: ubuntu-latest
    steps:
      - name: Install Python
        uses: actions/setup-python@v2
        with:
          python-version: 3.7
      - name: Checkout website source
        id: checkout-master
        uses: actions/checkout@v2
      - name: Cache dependencies
        uses: actions/cache@v2
        with:
          path: ~/.cache/pip
          key: pelican-4.2.0-${{ hashFiles('requirements.txt') }}
          restore-keys: |
            pelican-4.2.0-
      - name: Install Pelican
        run: |
          pip install -r requirements.txt
      - name: Run Pelican
        id: run-pelican
        run: |
          pelican content --output docs --settings publishconf.py
      - name: Push to GitHub pages
        id: push-to-gh-pages
        run: |
          git config --global user.name "my username"
          git config --global user.email "myemail@provider.com"

          git add docs/
          git commit --amend --no-edit

          git push origin master:deploy --force

Once you commit this file, a new Workflow should appear in GitHub Actions tab.

The detail important for this case is that once Pelican creates a docs/ folder, new changes are pushed to a new branch called deploy. I have configured this repository so that the static content is served from deploy branch, docs/ folder.

Every time I submit new commit(s), workflow is triggered (on push) and new contents delivered.

Pin package versions in your production Docker image

Roman Luštrik — Thu, 25 Jun 2020 12:34:00 +0200

Using package in R is easy. You install from CRAN using install.packages("packagename"), it resolves dependencies and you're good to go. What R natively doesn't handle so well is installing a particular package version without jumping through hoops. Technically you need the source file of the package version you want to install AND all source files of the dependencies (in the correct version, of course). This has been made almost seamless with packages packrat and recently, renv.

This comes handy when you are constructing a Docker file to run in production. Usually you want to run this defensively and do not want things to change from one image build to another. To get there, you can save all your package names and version into a file (renv.lock) and use that to reconstruct the now defined package structure with predictable versions (see renv vignette here).

For installing into Docker images, you can follow the official instructions in renv documentation. Below you'll find modifications I found necessary to get my packages installed.

My workflow is designed in such a way that I have a Dockerfile located in a folder above my R project I want to deploy. I install all the necessary tools and copy the application into the Docker image. One caveat I noticed when doing this, hidden files were copied too, and .Rprofile (more info i.e. here) was causing me a lot of grief. For example, it contained commands that were looking for renv/activate.R script, which was, naturally, non existent because that's not something you commit to your git repository. This was solved by overwriting it (or deleting it).

Here is an example that is working for me. See if you can find anything useful in it.

FROM rocker/r-ver:3.6.2

MAINTAINER Yours Truly "yours.truly@checksnbalances.com"

# Here is where I install all the necessary system libraries needed
# by R packages. Don't worry, R will, after compiling for 30
# minutes and file, tell you what packages you would need.
RUN apt-get update && apt-get install -y \
    zlib1g-dev \
    libcurl4-openssl-dev \
    libssl-dev

# This is something I use to deploy apps to shinyproxy. It is
# probably something that could be avoided by specifying host and
# port in `runApp()`. Note that the location may be OS dependent.
RUN echo 'local({options(shiny.port = 3838, shiny.host = "0.0.0.0")})' >> /usr/local/lib/R/etc/Rprofile.site

# Your code should not be run by root, so creating and switching to
# a new user. Feel free to come up with your own fun ID.
RUN useradd -m -u 2000 poldeta
USER poldeta

# Recursively make an R library folder. This is where installed R
# packages will be stored.
RUN mkdir /home/poldeta/R/library -p && mkdir /home/poldeta/shinyapp

# Moving to the app folder is probably not necessary at this
# particular point, but you know, whatever.
WORKDIR /home/poldeta/shinyapp

# Create .Rprofile site that will include your favorite (writable)
# location for installed R packages
RUN echo ".libPaths('/home/poldeta/R')" >> .Rprofile && R -e "install.packages('renv')"

# Switch to a superuser and copy your application into your Docker
# image.
USER root
COPY shinyapp /home/poldeta/shinyapp

# Make sure folder and file permissions are set to your new username.
RUN chown -R poldeta:poldeta /home/poldeta/shinyapp

# This is where the magic happens. When copying the app into the
# docker image, `renv.lock` file was also transferred. Because is
# being called from the working directory where `renv.lock` is
# located, calling `restore()` with defaults makes everything work
# as intended.
USER poldeta
RUN R -e "renv::restore()"

# Do any other necessary things to your image.
EXPOSE 3838

# Finally, run the app to be served.
CMD ["R", "-e", "shiny::runApp('/home/poldeta/shinyapp')"]

Odkrivanje elementov delovnega razmerja s strani Inšpektorata RS za delo

Roman Luštrik — Sat, 23 May 2020 10:48:00 +0200

Na Inšpektoratu RS za delo so bili tako prijazni, da so mi poslali stiskano in poskenirano tabelo zaključnenih dokumentov inšpekcijskih nadzorov, ki se tičejo ugotavljanja elementov delovnega razmerja kot jih opredeljuje ZDR-1. Po domače povedano so inšpektorji in inšpektorice pri delodajalcih preverjali, če ima podjetje delavce zaposlene "prek s.p.-ja".

Z inšpektorata so mi sporočili, da bi ugodenje zahtevi za vse dokumente, ki se tičejo ugotavljanja elementov delovnega razmerja, trajalo dolgo in bi bilo povezano s stroški. Čez to zgodbo že šel z Inšpektoratom RS za notranje zadeve (kurc enivan?), je Urad IP pokazal, da zakon ne predvideva, da se zaračunava stroške, če se zahteva zadevo elektronsko. Ker sem len, sem se odločil, da bom zahteval zadeve per partes. Tako so mi poslali lično natiskano in v PDF poskenirano tabelo relevantnih zadev.

Prejet PDF dokument sem spremenil v PDF s pomočjo paketa pypdfocr in s tesseract prebral številke zadev. Z manjšo skripto sem očistil datoteko. Vse skupaj je trajalo nekaj minut, jaz ne vem kaj si mislijo na Inšpektoratu.

import re
import random

with open('./data/seznam_zadev.txt', mode='rt') as rf:
    xy = rf.readlines()

out = []
for line in xy:
    mch = re.search('\d+-\d+', line)
    if mch:
        out.append(mch.group())

xy = []

for line in out:
    if line.startswith('0'):
        xy.append(line)
    else:
        xy.append(line[1:])

with open('./data/spucane_st_zadev.txt', mode='wt') as wf:
    for line in xy:
        wf.writelines(f'{line}\n')

random.seed(357)

random_rows = []
for rng in range(len(xy) - 1):
    random_rows.append(random.randrange(start=0, stop=len(xy)))

[xy[rng] for rng in random_rows]

Skripta vrne očiščeno datoteko in seznam zadev. Za te zadeve sem zaprosil in jih tudi dobil po dveh mesecih, verjetno zaradi koronast(r)anja.

'010601-201506'
'010606-201812'
'010505-201701'
'010104-201905'
'010611-201605'
'010601-201806'
'011401-201901'
'010506-202001'
'010401-201507'
'011305-201910'

Zadeva 06100-1427/2016

Pri podjetju iz primorske regije, ki se ukvarja s skladiščenjem in pretovarjanjem vozil, je inšpektorat ugotovil, da nima zaposlenih delavcev, na podjetju pa je bilo aktivnih šest delavcev. Ti niso imeli dovolilnice oz. niso zaposleni, imajo pa priglašeno dejavnost samostojnega podjetnika. Delodajalec je inšpektorju pokazal izjave, da delavci ne želijo biti zaposleni pri tem podjetju iz različnih razlogov. Ponudil je tudi dokaz, da jih je želel zaposliti, a te pogodbe delavci niso želeli skleniti. Tale navedba se mi je zdela zanimiva

ZDR-1 določa, da je potrebno skleniti pogodbo o zaposlitvi, svobodna in vzajemna volja ne more biti v nasprotju z zakonodajo.

Zadeva 710-2452/2018

Inšpektorica je ugotovila, da med podjetjem in samostojnim podjetnikom obstajajo elementi delovnega razmerja (v nasprotu z drugim odstavkom 13. člena ZDR-1) in izdala plačilni nalog za 1500 EUR globe. Sicer je zagrožena globa je do 8.000,00 EUR.

Zadeva 06100-1604/2016

Inšpektor z območne enote Celje je javnemu zavodu prepovedal opravljanje dela z delavcem, ki je delal kot samostojni podjetnik. Inšpektor je ugotovil, da obstajajo elementi delovnega razmerja, v ozir pa je vzel tudi pogodbo o izvajanju del.

Zadeva 710-1032/2019

Podjetje in podjetnik, ki padeta pod pristojnost inšpektorata iz Kranja sta dobila opomin, ker je bilo dovoljeno delavcu opravljati delo preko svoje družbe. To naj bi bilo v nasprotju s tretjo točko 217. člena ZDR-1, ki se glasi

pri njem opravlja delavec delo na podlagi pogodbe civilnega prava v nasprotju z drugim odstavkom 13. člena tega zakona;

Kršitelja morata poravnati stroške postopka v višini 30 EUR.

Zadeva 06109-7/2015

Murskosoboško-Mariborski inšpektorat je obiskal gostinsko podjetje. Delavka je izjavila, da ni zaposlena, ima s.p. in pogodbo o poslovnem sodelovanju. Inšpektorica je ugotovila, da obstajajo vsi elementi delovnega razmerja po 4. členu ZDR-1 (delavka se je prostovoljno vključena v organiziran delovni proces in s sredstvi delodajalca za plačilo po njegovih navodilih in njegovim nadzorom opravljala delo). Inšpektorica je prepovedala opravljanje dela delavke dokler je ne zaposli skladno s predpisi o zaposlovanju.

Zadeva 710-1666/2018-3

Kandidat je opravljal delo pooblaščene osebe v imenu za račun v korist in s sredstvi pravne osebe. Inšpektor je presodil, da gre za kršitev 4. člena ZDR-1 in izrekel opomin. Kršiteljici sta morali plačati 30 EUR stroškov postopka.

Zadeva 710-2628/2017

Podjetje s osebo z s.p.-jem sklenilo pogodbo za opravljanje dela direktorja v imenu in v korist pravne osebe z vsemi elementi delovnega razmerja. Inšpektorica je ugotovila, da gre za kršitev 13. člena ZDR-1. Pravni in odgovorni osebi je izrekla opomin in plačilo 30 EUR stroškov postopka.

Zadeva 710-3015/2019

Podobno kot za zadevo 710-2628/2017.

Zadeva 710-917/2019

Podobno kot za zadevo 710-2628/2017, le da je bila izrečena globa 1500 EUR za pravno in 150 EUR za odgovorno osebo. Dodatno so še zaračunali 180 in 70 EUR stroškov postopka za pravno in odgovorno osebo.

Zadeva 06141-1803/2019-5

Podjetju je inšpektorica prepovedala opravljanje dela na podlagi študentskih napotnic (strežba v lokalu). Inšpektorica je ugotovila, da študentke opravljajo delo v postrežbi v lokalu, ki vsebuje elemente delovnega razmerja. Delodajalcu je bilo naloženo, da v roku treh dni (od prejema odločbe) delavkam izročiti pisno pogodbo o zaposlitvi za delo v strežbi.

Naslednje zadeve sem prejel dne 2020-06-15

Zadeva PN 710-68/2019-3

Kratko a jedrnato poročilo in plačilni nalog za 1500 EUR oz 750 EUR za polovičko. Oglobljena je bila lastnica lokala, ker je v lokalu delala delavka neskladno z 2. odstavkom 13. člena Zakona o delovnih razmerjih, saj so obstajali elementi delovnega razmerja.

Zadeva 06100-1464/2018

Koprska inšpektorica je podjetju prepovedala opravljanje delovnega procesa oz. uporabo sredstev za delo s samostojnim podjetnikom do odprave nepravilnosti. Utemeljuje, da delodajalec omogoča opravljanje dela na podlagi pogodbe civilnega prava na način, da delo s.p.-ja vsebuje vse elemente delovnega razmerja. Inšpektorica odredi, da morata stranki skleniti pogodbo o zaposlitvi.

Zadeva 06100-2793/2019-7

Mariborska inšpektorica je delodajalcu prepovedala opravljati delo dvema osebama do odprave nepravilnosti. Delavca sta namreč opravljali delo na podlagi pogodbe civilnega prava, njuno delo pa vsebuje vse elemente delovnega razmerja. Ne piše (pobrisano), ali je šlo za fizični osebi ali morda za samostojna podjetnika.

Zanimivo mi je to, da je glava odločble na videz podobna tisti, ki jo je spisala koprska inšpektorica (ali pa je obratno, iz zaključnih zadev težko sklepam).

Zadeva 06141-576/2018

Inšpektor ugotavlja, da je imel delodajalec z dvema delavkama (s.p.?) sklenjeno pogodbo civilnega prava in da obstajajo elementi delovnega razmerja. Nalaga, da delodajalec uredi delovno razmerje skladno s 54. členom ZDR-1A.

Zadeva 06141/191/2015-1

Inšpektor delodajalcu prepove opravljanje dela oz delovnega procesa z delavko, s katero ima sklenjeno , dokler bo pri delodajalcu opravljala delo na način, ki vsebuje elemente delovnega razmerja in tako zagotovi spoštovanje določb 2. odstavka 13. člena ZDR-1.

Zadeva 710-2567/2017-3

Inšpektorica je izdala opomin in 30 EUR stroškov postopka, ker je podjetje dovolilo delavki opravljati delo na podlagi civilne pogodbe v nasprotju s 13. členom Zakona o delovnih razmerjih, ker vsebujejo elemente delovnega razmerja v skladu s 4. točko v povezavi z 22. členom ZDR-1.

Zadeva 06100-243/2016

Inšpektorica delodajalcu prepove opravljanje dela z osebo. Sicer je veliko pobrisanega, vendar bi lahko sklepal, da gre za delo na podlagi ("civilne") pogodbe, obstajajo pa tudi elementi delovnega razmerja skladno s 4. in v povezavi z 22. oz. 54. členom Zakona o delovnih razmerjih.

Zadeva 06139-1670/2016

Inšpektorica delodajalcu prepove opravljana dela oz. opravljanje delovnega procesa v katerem je udeležena delavka s katero ima, ker je pobrisano ugibam, sklenjeno pogodbo. Delo, ki ga delavka opravlja vsebuje vse elemente delovnega razmerja, kar je v nasprotju z 2. odstavkom 13. člena Zakona o delovnih razmerjih.

Zadeva 06100-157/2015-1

Inšpektor delodajalcu prepove opravljanje dela oz. delovnega procesa z delavcem, ki ima priglašeno dejavnost . Sklepam, da gre za delo preko s.p.-ja. Prepoved velja dokler delavce pri delodajalcu delo opravlja na način, ki vsebuje elemente delovnega razmerja.

Zadeva 710-332/2017

Inšpektor pravni in odgovorni osebi očita štiri prekrške od a do č. V dokumentu je nepobrisan samo prekršek a. V prekršku a je očitano, da je bilo delavcu omogočeno delo na podlagi pogodbe civilnega prava, kljub obstoju elementov delovnega razmerja, kar je v nasprotju z 2. odstavkom 13. člena ZDR-1. Pravna oseba je za ta prekršek dobila izrečene 4500 EUR globe, odgovorna pa 450 EUR. Poleg globe je še obračunana sodna taksa 1050 EUR za pravno in 160 za odgovorno osebo.

Naslednje zadeve sem prejel 2020-07-29

Zadeva 06139-392/2015-3

Inšpektor delodajalcu prepove opravljanje delovnega procesa z delavcem, ker mu omogoča opravljanje dela na podlagi civilnega prava - pogodbe o opravljanju storitev. Pogodba vsebuje elemente delovnega razmerja, v nasprotju z 2. drugim odstavkom 13. člena Zakona o delovnih razmerjih. Verjetno gre za s.p., a zaradi umika osebnih informacij iz tega dokumenta to nedvoumno ne znam razbrati. V primeru, da delodajalec ne bo odpravil nepravilnosti, je zagrožena globa 4500 EUR za pravno osebo in 350 EUR za odgovorno osebo.

06112-811/2014-1

Delavec s priglašeno dejavnostjo v gostinstvu na podlagi pogodbe o sodelovanju pri delodajalcu opravlja delo. Inšpektor je ugotovil, da obstajajo elementi delovnega razmerja po drugem odstavku 13. člena Zakona o delovnih razmerjih. V kolikor delodajalec ne odpravi nepravilnosti, je zagrožena globa 4500 EUR za pravno in 350 EUR za odgovorno osebo.

06141-590/2016

Inšpektorica je dvema delavkama s priglašeno dejavnostjo prepovedala opravljanje dela pri delodajalcu, ker je slednji omogočil opravljanje dela na podlagi pogodb civilnega prava, kar je v nasprotju z drugim odstavkom 13. člena Zakona o delovnih razmerjih, saj vsebuje delo vse elemente delovnega razmerja. Zagrožena globa je 4500 EUR za delodajalca in 350 EUR za odgovorno osebo.

06100-583/2016

Inšpektor delodajalcu prepove opravljati dela oz. opravljanje delovnega proces in uporabe sredstev za delo z delavko, ki ima priglašeno dejavnost dokler bo pri delodajalcu opravljala delo na način, ki vsebuje elemente delovnega razmerja, kar je v nasprotju z drugim odstavkom 13. člena Zakona o delovnih razmerjih. Zagrožena je kaze 4500 EUR za delodajalca in 350 EUR za odgovorno osebo.

06100-1228/2016

Inšpektor je ob inšpekcijskem pregledu ugotovil, da imajo nekateri delavci status (pobrisano, verjetno s.p.), z družbo sklenjeno pogodbo civilnega prava. Inšpektor je ugotovil, da obstajaji vsi elementi delovnega razmerja, kot jih določa 4. člen Zakona o delovnih razmerjih-1 in ni v skladju z drugim odstavkom 13. člena istega zakona. Delodajalec je obljubil, da bo z navedenimi zaposlenimi sklenil pogodbo o zaposlitvi do 2017-01-01, ker potrebuje nekaj časa za spremembo internega akta. Za delodajalca je tudi zagroženakazen 2000 EUR in 200 EUR za neizpolnjevanje pisnega poročila ter 4500 EUR kazni za neizvrševanje določbe za pravno in 350 EUR za odgovorno osebo.

710-603/2019-3

V tej zadevi je veliko informacij umaknjenih in je težko razbrati za kaj gre. Da se razbrati, da je pri pravni osebi nekdo opravljal dleo na podlagi pogodbe civilnega prava s čimer je kršila drugi odstavek 13. člena Zakona o delovnih razmerjih-1. Pravni in odgovorni osebi je bil izrečen opomin in vsaki tudi plačilo sodne takse v višini 30 EUR.

710-2691/2017-3

Inšpektorica je ugotovila, da je bila kršiteljica naročnica del z brezposelno osebo. Ugotovljeni so bili elementi delovnega razmerja, kar je v nasprotju z drugim odstavkom 13. člena Zakona o delovnih razmerjih-1 in je tako storila prekršek po tretji točki prvega odstavka v povezavi z drugim odstavkom 217. člena Zakona o delovnih razmerjih [ja, taka klobasa v odločbi piše]. Izrečen ji je bil opomin in plačilo stroške postopka odmerjene sodne takse v vrednosti 30 EUR.

710-2214/2018

Inšpektor je ugotovil, da je kršiteljica omogočila/dovolila, da je pri njej na podlagi pogodbe civilnega prava opravljala delo oseba (pobrisano), kar je v nasprotju z določbami 2. odstavka 13. člena Zakona o delovnih razmerjih-1. Samostojni podjetnici, ki zaposluje manj kot 10 oseb, se tako izreče globa 1500 EUR in 150 EUR sodne takse.

710-1885/2017

Inšpektorica pravni in odgovorni osebi izreče opomin, ker je omogočila opravljane dela prodajalke, verjetno na podlagi pogodbe civilnega prava, na sistematiziranem delovnem mestu prodajalke z vsemi elementi delovnega razmerja. To je v nasprotju z drugim odstavkom 13. člena Zakona o delovnih razmerjih. Pravna in odgovorna oseba morata vsaka plačati 30 EUR stroškov postopka - sodno takso.

710-1390/2017

Inšpektor pravni in odgovorni osebi očita odgovornost za prekšek, ker je odgovorna oseba v imenu pravne osebe sklenila pogodbo civilnega prava z osebo, ki ima priglašeno dejavnost, in ji tako omogočila delo, ki je v nasprotju z drugim odstavkom 13. člena Zakona o delovnih razmerjih. Izrečena je globa 3000 EUR za pravno in 450 EUR za odgovorno osebo.

Naslednje zadeve sem prejel 2020-09-29

06100-1223/2018-7

V gostinskem lokalu je inšpektorica odredila prepoved opravljanja dela delavcev (5) do odprave nepravilnosti. Odprava izgleda tako, da bo podjetje moralo, v kolikor želi sodelovati s temi petimi osebami, zaposliti, saj vsebina dela vsebuje vse elemente delovnega razmerja. Inšpektorica je ugotovila tudi druge kršitve, denimo opravljanje dela na podlagi podjemne pogodbe, kjer delavec ni bil zavarovan za primer poklicne bolezni in primer poškodbe pri delu.

710-243/2018-3

Pravna oseba in njena odgovorna oseba (direktor) sta sklenila pogodbi o sodelovanju, v katerih obstajajo elementi delovnega razmerja v skladu s 4. in v povezavi z 22. oz 54. členom, kar je v nasprotju z 2. odstavkom 13. člena ZDR1. Pravna in odgovorna oseba dobita opomin in 30 EUR kazni (vsaka).

710-635/2019-3

Kršitelj je prejel opomin in plačilo sodne takse (30 EUR), ker je omogočil natakarici delo na podlagi podjemne pogodbe. To delo ima vse elemente delovnega razmerja v skladu s 4. in v povezavi z 22. oz. 54. čl. Zakona o delovnih razmerjih (Ur.l.RS, štev. 21/2013 in 52/2016) [si že videl/a toliko pik v eni povedi?].

06139-831/2016

Pravni osebi se prepove opravljanje dela voznika delavcu dokler se ne odpravi nepravilnosti. Inšpektor očita nepravilnost, ki je bila storjena s tem, da je delavcu delodajalec omogočil opravljanje dela sistematiziranega delovnega mesta z vsemi elementi delovnega razmerja v nasprotju z 2. odstavkom 13. člena ZDR. To nepravilnost so odkrili sicer pri nadzoru nad opravljanjem dela dijakov in študentov z napotnico študentskega servisa.

06100-2128/2017-3

Delodajalcu se do odprave nepravilnosti prepove opravljanje dela delavke, ker je dopustil opravljanje dela (v kuhinji) na podlagi civilne pogodbe v nasprotju s 13. členom ZDR-1, in sicer zato, ker vsebuje elemente delovnega razmerja v skladu s 4. točko v povezavi z 22. členom istega zakona. V tej odločbi je inšpektorica opozorila, da v primeru, da bo dvakrat v šestih mesecih ali več kot dvakrat v daljšem obdobju ponovila kršitev, bo zavezancu zapečatila sredstva za delo, delovne prostore ter delovišče ter druge prostore.

06100-1020/2017-011003

Delodajalcu se prepove opravljanje delovnega procesa oziroma uporabe sredstev za delo z dvema delavcema (špedicija in druge spremljajoče prometne dejavnosti). Nepravilnost je storjena s tem, da jima je omogočil opravljanje dela na podlagi pogodb civlnega prava na način, da njuno delo vsebuje vse elemente delovnega razmerja. Prepoved velja do odprave nepravilnosti.

06100-1427/2016

Delodajalcu se prepove opravljanje dela oziroma opravljanje delovnega procesa in uporabe sredstev s šestimi delavci, dokler bodo pri delodajalcu opravljali delo na način, ki vsebuje elemente delovnega razmerja in s tem zagotovi spoštovanje določb 2. odstavka 13. člena ZDR-1.

710-985/2019-4

Kršiteljica je kot odgovorna oseba z drugo osebo sklenila pogodbo za opravljanje dela. S tem je kršila 2. odstavek 13. člena ZDR, ker obstajajo elementi delovnega razmerja v skladu s 4. in v povezavi z 22. oz. 54. členom Zakona o delovnih razmerjih. Izrečen je bil opomin in plačilo stroške postopka v višini 30 EUR.

710-1495/2018-2

V delavnici strojnega varjenja filtrov, valjanja in brušenja je ob inšpekcijskem nadzoru pri zavezancu inšpektorica odkrila, da pri zavezancu opravlja delo oseba, kjer so vsi elementi delovnega razmerja v skladu s 4. in v povezavi z 22. členom ZDR-1. Pravni osebi je naložila 3.000,00 EUR, odgovorni pa 450,00 EUR kazni (z možnostjo plačile polovičnega zneska).

06139-869/2015

Delodajalec more dvema delovkama (pomoč pri postrežbi pijač in čiščenje lokala) skleniti pogodbo o zaposlovanju in ju prijaviti v zavarovanje. S tem bo zagotovil spoštovanje 2. odstavka 13. člena ZDR-1.

Docker image for OBITools (1.2.13) available

Roman Luštrik — Sat, 15 Jun 2019 00:00:00 +0200

In order to aid in making bioinformatics pipelines more reproducible, I have put together a docker image which houses OBITools. OBITools is a software bundle which helps you analyze NGS data. See their website for more information.

Having obitools housed in a docker image enables us to easily migrate or communicate our workflow. Here is a short representation of how one can call functions from the bundle using Python's docker package.

First, prepare some boilerplate stuff to have the container ready as a Python variable.

import docker
from docker.errors import NotFound

client = docker.from_env()

# Create or start docker container named 'obitools'.
try:
    obt = client.containers.get('obitools')
    if obt.status == 'exited':
        print('Starting container obitools.')
        obt.start()
    else:
        print('Container obitools ready.')
except NotFound:
    print('Creating container obitools.')
    obt = client.containers.run(
        image='romunov/obitools:1.2.13',
        name='obitools',
        # This maps my local wolfdata to /data in container
        volumes={'/home/romunov/wolfdata': {'bind': '/data', 'mode': 'rw'}},
        tty=True, detach=True
        )

And then run the command

score_min = 40
reads_r = "wolf_R.fastq"
reads_f = "wolf_F.fastq"
output = "wolf.fastq"
wolf = obt.exec_run(tty=True, workdir='/data', cmd=f'illuminapairedend --score-min={score_min} -r {reads_r} {reads_f} > {output}')

Command illuminapairedend --score-min=40 -r wolf_R.fastq wolf_F.fastq > wolf.fastq would be sent to the docker container and output would be located in /home/romunov/wolfdata (see the first code chunk, specifically study the argument volumes).

Insert changes into deep commits

Roman Luštrik — Fri, 14 Jun 2019 00:00:00 +0200

Have you ever committed a branch and review addressed issues from all commits? This post will address the issue of making changes at the top of the commit tree and then committing them deep to their respective commits.

A snapshot of the situation is illustrated in the below code chunk.

romunov@mucek:~/Documents/fixme$ git log
commit 1f78688a40598552262af08598fb4eb256421df8 (HEAD -> master)
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:47 2019 +0200

    Add file3

commit f7defe93b96436047988e9e5d943f22c9e136ad4
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:34 2019 +0200

    Add file2

commit 27a395f1f8878d83b8fc03629f1ef455f56db616
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:16 2019 +0200

    Add file1

commit 3fa086c62d4cf0041169a50946e0cc428fcc90d8
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:21:55 2019 +0200

    welcome

romunov@mucek:~/Documents/fixme$ git status
On branch master
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   file1.txt
    modified:   file2.txt
    modified:   file3.txt

no changes added to commit (use "git add" and/or "git commit -a")

I will show you how to iteratively make commits of files, hiding the rest of the uncommited changes, pushing those added files to the appropriate commit and then rinse and repeat until all the changes are done.

Starting from scratch, let's create a folder called fixme that will contain 4 files. Each file will reside in its own commit.

romunov@mucek:~/Documents$ mkdir fixme
romunov@mucek:~/Documents$ cd fixme/
romunov@mucek:~/Documents/fixme$ git init
Initialized empty Git repository in /home/romunov/Documents/fixme/.git/

romunov@mucek:~/Documents/fixme$ echo "welcome" > welcome.txt
romunov@mucek:~/Documents/fixme$ git add welcome.txt
romunov@mucek:~/Documents/fixme$ git commit -m "welcome"
[master (root-commit) 3fa086c] welcome
 1 file changed, 1 insertion(+)
 create mode 100644 welcome.txt

romunov@mucek:~/Documents/fixme$ echo "txt in file1" > file1.txt
romunov@mucek:~/Documents/fixme$ git add file1.txt
romunov@mucek:~/Documents/fixme$ git commit -m "Add file1"
[master 27a395f] Add file1
 1 file changed, 1 insertion(+)
 create mode 100644 file1.txt

romunov@mucek:~/Documents/fixme$ echo "txt in file2" > file2.txt
romunov@mucek:~/Documents/fixme$ git add file2.txt
romunov@mucek:~/Documents/fixme$ git commit -m "Add file2"
[master f7defe9] Add file2
 1 file changed, 1 insertion(+)
 create mode 100644 file2.txt

romunov@mucek:~/Documents/fixme$ echo "txt in file3" > file3.txt
romunov@mucek:~/Documents/fixme$ git add file3.txt
romunov@mucek:~/Documents/fixme$ git commit -m "Add file3"
[master 1f78688] Add file3
 1 file changed, 1 insertion(+)
 create mode 100644 file3.txt

romunov@mucek:~/Documents/fixme$ git status
On branch master
nothing to commit, working tree clean

romunov@mucek:~/Documents/fixme$ git log
commit 1f78688a40598552262af08598fb4eb256421df8 (HEAD -> master)
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:47 2019 +0200

    Add file3

commit f7defe93b96436047988e9e5d943f22c9e136ad4
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:34 2019 +0200

    Add file2

commit 27a395f1f8878d83b8fc03629f1ef455f56db616
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:16 2019 +0200

    Add file1

commit 3fa086c62d4cf0041169a50946e0cc428fcc90d8
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:21:55 2019 +0200

    welcome

Now let's create some changes to all files.

romunov@mucek:~/Documents/fixme$ echo "another txt in file 1" >> file1.txt
romunov@mucek:~/Documents/fixme$ echo "another txt in file 2" >> file2.txt
romunov@mucek:~/Documents/fixme$ echo "another txt in file 3" >> file3.txt
romunov@mucek:~/Documents/fixme$ git status
On branch master
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   file1.txt
    modified:   file2.txt
    modified:   file3.txt

no changes added to commit (use "git add" and/or "git commit -a")

Let's push each change for file1, file2 and file3 to its corresponding commit. The steps taken are 1. add specific file 2. push to specific commit using --fixup 3. hide uncommited changes using git stash 4. rebase 5. recall uncommited changes using git stash apply and go from 1.

Some comments about the code are inline.

romunov@mucek:~/Documents/fixme$ git add file1.txt
# To get hash of a commit, see above or use `git log`.
romunov@mucek:~/Documents/fixme$ git commit --fixup=27a395f1f8878d83b8fc03629f1ef455f56db616
[master f24f11b] fixup! Add file1
 1 file changed, 1 insertion(+)

 # Hide uncommited changes.
romunov@mucek:~/Documents/fixme$ git stash
Saved working directory and index state WIP on master: f24f11b fixup! Add file1

# Notice that the new fixedup commit is on top.
romunov@mucek:~/Documents/fixme$ git log
commit f24f11b7ca983be10b2cf3f477ebd649542d3ba2 (HEAD -> master)
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:24:14 2019 +0200

    fixup! Add file1

commit 1f78688a40598552262af08598fb4eb256421df8
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:47 2019 +0200

    Add file3

commit f7defe93b96436047988e9e5d943f22c9e136ad4
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:34 2019 +0200

    Add file2

commit 27a395f1f8878d83b8fc03629f1ef455f56db616
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:16 2019 +0200

    Add file1

commit 3fa086c62d4cf0041169a50946e0cc428fcc90d8
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:21:55 2019 +0200

    welcome

# Rebase. Make sure you go back in time (HEAD~X) so that you cover them
# actual commit message. Use autosquash so you don't have to manually
# squash it in the next window.    
romunov@mucek:~/Documents/fixme$ git rebase -i HEAD~4 --autosquash

Successfully rebased and updated refs/heads/master.

# The top "fixup" commit has been squashed with commit titled
# "Add file1".
romunov@mucek:~/Documents/fixme$ git log
commit 95e48ac32aa93256aa248f8077c293bf16b8d911 (HEAD -> master)
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:47 2019 +0200

    Add file3

commit 7e881feb91bdf326a3af985697d7837283a2ba12
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:34 2019 +0200

    Add file2

commit 6d52f8818d587fded3910dd688f4fe6a762fd7dd
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:22:16 2019 +0200

    Add file1

commit 3fa086c62d4cf0041169a50946e0cc428fcc90d8
Author: Roman Luštrik <romunov@gmail.com>
Date:   Thu Jun 13 16:21:55 2019 +0200

    welcome

We now bring back uncommited changes...

romunov@mucek:~/Documents/fixme$ git stash apply
On branch master
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   file2.txt
    modified:   file3.txt

no changes added to commit (use "git add" and/or "git commit -a")

... and go through the same process again.

romunov@mucek:~/Documents/fixme$ git add file2.txt
romunov@mucek:~/Documents/fixme$ git commit --fixup=7e881feb91bdf326a3af985697d7837283a2ba12
[master d3f6227] fixup! Add file2
 1 file changed, 1 insertion(+)

romunov@mucek:~/Documents/fixme$ git stash
Saved working directory and index state WIP on master: d3f6227 fixup! Add file2
romunov@mucek:~/Documents/fixme$ git rebase -i HEAD~4 --autosquash

Successfully rebased and updated refs/heads/master.

The final file should go into the top commit, so we can just use git commit --amend.

# Bring up uncommited changes.
romunov@mucek:~/Documents/fixme$ git stash apply
On branch master
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   file3.txt

no changes added to commit (use "git add" and/or "git commit -a")
romunov@mucek:~/Documents/fixme$ git add file3.txt
romunov@mucek:~/Documents/fixme$ git commit --amend
[master c251243] Add file3
 Date: Thu Jun 13 16:22:47 2019 +0200
 1 file changed, 2 insertions(+)
 create mode 100644 file3.txt

You should now have a clean workspace with changes in their respective commits.

CAVEAT Do this only on private branches, unless you are the only person working on master branch. If you force push these changes, your collaborators may throw you off of the nearest bridge.

Vectorizing functions in R is easy

Roman Luštrik — Wed, 17 Apr 2019 00:00:00 +0200

Imagine you have a function that only takes one argument, but you would really like to work on a vector of values. A short example on how function Vectorize() can accomplish this. Let's say we have a data.frame

xy <- data.frame(sample = c("C_pre_sample1", "C_post_sample1", "T_pre_sample2",
                            "T_post_sample2", "NA_pre_sample1"),
                 value = runif(5))

#           sample     value
# 1  C_pre_sample1 0.3048032
# 2 C_post_sample1 0.3487163
# 3  T_pre_sample2 0.3359707
# 4 T_post_sample2 0.6698358
# 5 NA_pre_sample1 0.9490707

and you want to subset only samples that start with C_pre or T_pre. Of course you can construct a nice regular expression, implement an anonymouse function using lapply/sapply or use one of those fancy tidyverse functions.

A long winded way would be to find matches using regular expression for each level, combine them and subset. This is for pedagogical reasons, so please bare with me.

i.ind <- do.call(cbind, list(
  grepl(pattern = "^C_pre", x = xy$sample),
  grepl(pattern = "^T_pre", x = xy$sample)
))

i.ind
#       [,1]  [,2]
# [1,]  TRUE FALSE
# [2,] FALSE FALSE
# [3,] FALSE  TRUE
# [4,] FALSE FALSE
# [5,] FALSE FALSE

# Find those rows in `xy` that have at least one TRUE and use that to subset the
# data.frame.
xy[rowSums(i.ind) > 0, ]

#          sample     value
# 1 C_pre_sample1 0.3048032
# 3 T_pre_sample2 0.3359707

The same can be achieved using a vectorized version of the grepl function. We designate which argument exactly is being vectorized, in our case pattern because that's the argument that is varying.

vgrepl <- Vectorize(grepl, vectorize.args = "pattern")

Here we use function Vectorize and we tell it to vectorize argument pattern. What this will do is run the grepl function for any element of the vector we pass in, just like we did in the i.ind objects a few lines above.

This would be an equivalent of doing it using an anonymouse function

tmp <- sapply(c("^C_pre", "^T_pre"), FUN = function(pt, input) {
  grepl(pt, x = input)
}, input = xy$sample)

tmp
#      ^C_pre ^T_pre
# [1,]   TRUE  FALSE
# [2,]  FALSE  FALSE
# [3,]  FALSE   TRUE
# [4,]  FALSE  FALSE
# [5,]  FALSE  FALSE

While this can be somewhat verbose, you can use vgrepl as you would use grepl, with the minor detail that you pass a whole vector to pattern instead of a single regular expression.

i.vec <- vgrepl(pattern = c("^C_pre", "^T_pre"), x = xy$sample)
#      ^C_pre ^T_pre
# [1,]   TRUE  FALSE
# [2,]  FALSE  FALSE
# [3,]  FALSE   TRUE
# [4,]  FALSE  FALSE
# [5,]  FALSE  FALSE

xy[rowSums(i.vec) > 0, ]

#          sample     value
# 1 C_pre_sample1 0.3048032
# 3 T_pre_sample2 0.3359707

How I implemented googleSignIn in R (shiny) and lived

Roman Luštrik — Sat, 16 Mar 2019 00:00:00 +0100

Known user identity when building shiny apps can sometimes come really handy. While you can implement your own user login, for instance using cookies, you can also use some of the services which authenticate a user for you, such as Google. This way, you don't have to handle cookies or passwords, just a small part of bureaucracy in your database.

Enter googleAuthR. You can use the Google Authentication system, which enables you to call its APIs (e.g. see here), but sometimes a user login is just enough. This is where the accompanying module googleSignIn comes into play. Below is how I was able to implement this same name module with four lines of code, one line being loading the aforementioned library.

If you would be kind enough to scroll to the googleSignIn part of the page, you will notice code for a shiny app that (almost) just works. Read on to see what I mean by "almost".

The demo app has all the key components, it loads libraries, prepares options("client_id"), creates sign in button in ui (googleSignInUI("demo")), does the auth dance for you in the server part (shiny::callModule(googleSignIn, "demo")) when this button is pressed and, as gratis, displays the name, email and your google image. I run the app using runApp(port = 1221)!

When I press the Sign in button, I get - bleh. Doesn't work.

What is going on? The google console parameters check out.

Let me save you hours of clicking by pointing out this little bit from the documentation from the googleAuth module:

... then make sure if you launch your app locally to change the ip address from 127.0.0.1 to localhost in your browser (Google doesn’t accept ip addresses).

In other words, in your web browser, just change http://127.0.0.1:1221 to http://localhost:1221. The page should reload on its own. Click the sign in button et voilà.

How to authenticate using OAuth2 through R

Roman Luštrik — Sun, 20 Jan 2019 00:00:00 +0100

If you need to have authentication of users in your application, you could invent the proverbial warm water by implementing register, login, logout and other features. Or, you could outsource part of that functionality to well established establishments such as Google, Facebook, Github and other. In addition to knowing the identity of your user, you can potentially gain access to service's APIs, which can be very handy (such as access to users' calendars, e-mail boxes, etc...).

In this post, you will learn how to set up authentication using OAuth2 using google, but the process is similar(ly painful) for other services.

If you do not know how OAuth2 works, you may want to check out the figures here and here. Basically it works by sending your user (pops up a website) to a service (e.g. Google), where they confirm access to their data (scope). On return, they will bring with them a "code". This code is then traded in by you at the service desk for a token, which acts as a pass to see particular user's data. Store this somewhere safe.

To accomplish this, R package httr comes equipped with all the lingo needed to successfully talk to services. There are some good demos in package source code, however, it perhaps lacks some minor details. This post will perhaps shine some light on those details, but who knows how long the screenshots will be relevant. Hopefully at least ideas will be evergreen for the foreseeable future.

If you haven't done so yet, head over to the google developers' console and create a new project. In the Credentials menu, create new OAuth client ID credentials.

Select Web application and fill out application name and Authorized redirect URIs. This is the URI where user will be diverted to once the authentication has been confirmed. For testing purposes, make sure this is http://localhost/ (notice the trailing slash). Once deployed, the URI here would be sent back to your application's address. You can fetch secret and key from the Client ID for Web application menu, depicted as black stripes on the figure below.

Make sure you visit the OAuth consent screen tab and fill in any necessary information needed for transparent functioning of your application.

# Below are mock secret and key, they will not work. They are just an example of what they
#  would look like. Replace with your.
secret <- "vxR9AuyJ4cEwrPNaylaTyC3AfXWIEdQnFotju9Yc6Q4og.apps.googleusercontent.com
key <- "afL4IguOXCXC6-bPV9lObvxT"

We send the user to choose which account to authenticate with (a new window pops open) and then are returned a token. This token is then used to request any information needed by your app and authorized by the user. To read more on Google scopes, read here.

This is an example of scope for OAuth2 API.

# Ask user to authorize access to his or her data and gain a token.
auth.code <- oauth2.0_token(endpoint = oauth_endpoints("google"),
                            app = app,
                            scope = "https://www.googleapis.com/auth/userinfo.profile")

# Use token to fetch the actual data.
req <- GET("https://www.googleapis.com/oauth2/v1/userinfo",
           config(token = auth.code))

# This step is needed so that if the call fails, you will be
#  able to detect it and handle it R-way. You are best off wrapping this into
#  a `tryCatch` call if not in interactive session.
stop_for_status(req)

The result would look something along these lines.

# I changed sensitive information for this display, so don't bother. :)
> str(content(req))
List of 8
 $ id         : chr "754846239037474839489"
 $ name       : chr "Roman Luštrik"
 $ given_name : chr "Roman"
 $ family_name: chr "Luštrik"
 $ link       : chr "https://plus.google.com/754846239037474839489"
 $ picture    : chr "https://lh500600.googleusercontent.com/-32gefACRE/ACCAXSDFCASC/XAAXEXF/asdCac_/photo.jpg"
 $ gender     : chr "other"
 $ locale     : chr "en"

If you would also like to fetch user's e-mail address, you will need to extend the scope. Note that the query has changed and you will be, once again, asked to confirm that you allow your e-mail address be accessed by the app.

auth.code <- oauth2.0_token(endpoint = oauth_endpoints("google"),
                            app = app,
                            scope = c("https://www.googleapis.com/auth/userinfo.profile",
                                      "https://www.googleapis.com/auth/userinfo.email")
)

req <- GET("https://www.googleapis.com/oauth2/v1/userinfo",
           config(token = auth.code))

stop_for_status(req)
str(content(req))

List of 10
 $ id           : chr "754846239037474839489"
$ email         : chr "name@gmail.com"
$ verified_email: logi TRUE
$ name          : chr "Roman Luštrik"
$ given_name    : chr "Roman"
$ family_name   : chr "Luštrik"
$ link          : chr "https://plus.google.com/754846239037474839489"
$ picture       : chr "https://lh500600.googleusercontent.com/-32gefACRE/ACCAXSDFCASC/XAAXEXF
$ gender        : chr "other"
$ locale        : chr "en"

Error log

Restarting my R session, updating httr and making sure that the redirect URI was set to http://localhost/ solved my problem with this error below.

> ggl.token <- oauth2.0_token(endpoint = oauth_endpoints("google"),
+                             app = app,
+                             scope = "https://www.googleapis.com/auth/userinfo.profile")
Waiting for authentication in browser...
Press Esc/Ctrl + C to abort
Authentication complete.
Error in curl::curl_fetch_memory(url, handle = handle) :
  Could not resolve host: accounts.google.com

I came to the idea of restarting my R session when surefire stuff didn't work, such as updating R package(s) using devtools::install_github.

> devtools::install_github("r-lib/httr")
Installation failed: curl::curl_fetch_disk(url, x$path, handle = handle) : Could not resolve host: raw.githubusercontent.com

How to combine plots and table made with ggplot (or grid graphics) in R

Roman Luštrik — Tue, 02 Oct 2018 00:00:00 +0200

In the last few years, a number of options of how to combine grid graphics (incl. ggplot2) have emerged. If you can't remember all functions to do this, this post may serve as a reference guide, but it's mostly for me because I keep forgetting the functionalities provided by cowplot and patchwork packages (thank you Jaap). Here appearing in order I encountered them.

library(ggplot2)

# prepare some figures
fig1 <- ggplot(mtcars, aes(x = mpg, y = disp, color = as.factor(gear))) +
  theme_bw() +
  theme(legend.position = "top") +
  scale_color_discrete(name = "gear") +
  geom_point()

fig2 <- ggplot(mtcars, aes(x = gear, y = disp, fill = as.factor(gear))) +
  theme_bw() +
  theme(legend.position = "top") +
  scale_fill_discrete(name = "gear") +
  geom_violin()

gridExtra

Here is oldie but goldie from Baptiste's gridExtra package. You can, for example, specify the layout matrix or specify number of columns. The below code chunk is using ggplot2::ggsave which saves the last subplot only, which is why I saved the result of grid.arrange into a new variable (but see cowplot below).

library(gridExtra)

fig.combined1 <- grid.arrange(fig1, fig2, ncol = 2)
ggsave(plot = fig.combined1, filename = "grid.arrange.png",
       width = 8, height = 4, units = "in", dpi = 96)

cowplot

Next is cowplot, which comes with nice vignettes. What I like about this package is the elegant adding of letters to subplots, e.g.

library(cowplot)

plot_grid(fig1, fig2, labels = "AUTO")
cowplot::ggsave(filename = "plotgrid.png", width = 8,
                height = 4, units = "in", dpi = 96)

See the vignettes for more information. Package also packs its own ggplot which masks the function ggplot2::ggsave (read: ggsave from package ggplot2). The beauty of it is that it saves the gridded plot, not the last (sub)plot.

patchwork

Thomas did a great job of making combining of plots trivially easy. The GitHub repository of the package is a rich source of information and I could never do it justice. Here's my botched attempt of just an example of what the package functionality provides.

library(patchwork)

fig1 + fig2
cowplot

Point arrows to specific parts of the data

Roman Luštrik — Sat, 14 Apr 2018 00:00:00 +0200

tl;dr: Point arrows at your line at defined datum points.

Sometimes to really enhance your picture you need to point at some stuff. Here is a script which will plot vertical lines to show events in your datum plot.

library(ggplot2)

# This script will draw points to parts of (interpolated) lines.

arrow.length <- 10
touchoff.distance <- 4 # distance between data and start of arrow
arrowhead.size <- 2.5 # in millimeters

# Simulate some data.
set.seed(12)
datum.seq <- seq(from = as.Date("2018-01-01"),
                 to = as.Date("2018-12-31"),
                 by = "2 week")

xy <- data.frame(datum = rep(datum.seq, times = 2))
xy$count <- c(runif(length(datum.seq), min = 20, max = 50),
              runif(length(datum.seq), min = 40, max = 60)
)
xy$var1 <- rep(c("s1", "s2"), each = length(datum.seq))

# These are the points where we want to point to.
topoint <- data.frame(datum = sample(datum.seq, 16) - 7,
                      var1 = rep(c("s1", "s2"), each = 8))

# Function which interpolates per one variable.
interpolateOnDate <- function(xy, newdata) {
  newdata <- newdata[newdata$var1 == unique(xy$var), ]
  out <- approx(x = xy$datum, y = xy$count, xout = newdata$datum)
  newdata$count <- out$y
  newdata
}

# Apply interpolation to each level of var1
res <- lapply(split(xy, f = xy$var1),
              FUN = interpolateOnDate,
              newdata = topoint)

res <- do.call(rbind, res) # merge result

ggplot(xy, aes(x = datum, y = count)) +
  theme_bw() + theme(axis.title.y = element_blank()) +
  geom_point(shape = 1) +
  geom_path() +
  geom_segment(data = res, aes(x = datum, y = count + touchoff.distance,
                               xend = datum, yend = count + touchoff.distance + arrow.length),
               arrow = arrow(length = unit(arrowhead.size, "mm"), ends = "first")) +
  facet_wrap(~ var1, ncol = 1)

How to build a static website using Pelican

Roman Luštrik — Mon, 19 Mar 2018 00:00:00 +0100

It's fairly easy to build a static website. Static websites differ from dynamic in that they do not communicate with a database of any sorts. This makes it easier and especially cheaper to deploy to almost any setup which runs at least rudimentary serving capabilities.

There are many options for static site generation, I chose Pelican. There is no particular adamant reason behind it, but browsing for themes did tilt the scales in its favor.

Workflow is basically simple. One time install is that of Python (e.g. version 3.6). Once you have that, I would recommend you also get pip install virtualenvwrapper.

You then activate virtual environment and install all dependencies for pelican.

# create and activate virtual environment
mkvirtenv mystaticwebsite
workon mystaticwebsite

# installs pelican
pip install pelican

Create your project folder

mkdir mywebsite
cd mywebsite

Now feel free to follow instructions from pelican-quickstart on.

Once you are done writing contents to a markdown (.md) file you 'publish' your website, which puts all the bits and pieces together (applies settings, theme...) into an /output folder. You can do that using pelican content --output output --ignore-cache --relative-urls --settings publishconf.py. Making paths relative enables you to view your site locally without resorting to changing SITEURL = '' before compiling.

To make your website ready to be published, feel free to include settings specified in the publishconf.py, e.g. pelican content --output output --ignore-cache --settings publishconf.py. Ignoring cache will make sure some weird bits do not get left in and Pelican starts afresh.