Repository navigation
Create a Github Action version of the Bitbucket pipeline #65
Description
The current pipeline is meant to be run on Bitbucket CI. An equivalent Github CI version should be incorporated as well, and maintained. @floswald might be interested.
Metadata
Metadata
Assignees
Labels
No labels
Activity
floswald commentedon Sep 3, 2025
yes true, I could be interested. To be frank I created a julia package which I run locally (for now - on gh actions soon) in order to prepare some metrics about each package (most of it more or less freely copied from your scripts on the bitbucket pipeline). Main reason is that I'm better with julia than with bash; one advantage of your script-based pipeline is that bash is of course pre-installed on any runner, so no extra time spent installing julia and other latency. I have to say that this is quite fast nowadays, but still. the advantage of the julia package for me is that I find it easier to maintain and test. it's not well documented, but here is the main function that runs a few of your checks
yes the package is called https://github.com/JPE-Reproducibility/JPEtools.jl but of course that's only for lack of a better name and there is nothing JPE specific.
larsvilhuber commentedon Sep 3, 2025
I get why you use Julia, but using a statistical language to do raw text processing (and running Julia from command line to have Julia call out to the command line... ) is not my favorite.
That being said, I'm primarily interested in getting this into a more general usage, and both of our unfamiliarity with Python may be the golden middle (way more CS students who know Python than Julia).
Let's see what we can put together, presumably moving part of this to the SSDE github (neither AEA nor JPE...)
floswald commentedon Sep 3, 2025
I wouldn't brush it off as "a statistical language" tbh - I find it as general purpose as python (and often more so). It's really good for text manipulation, contrary to what you seem to think ;-)