update_repo: Run 2x to alleviate errors

[joeflack4]

Overview

I'm not sure why, but update_repo currently produces errors the first time you run it (after any changes): monarch-initiative/mondo-ingest#308

Until that gets fixed, the current practice is that users must run update_repo a second time. I propose that rather than requiring users to know that they need to do this and remember to do it, we automatically run update_repo 2x in a row. See: ignoring errors in make

[matentzn]

I thought about this problem for a bit now, also with ChatGPT, and some braincells active (at least 6 of my 23 braincells). I cannot figure out how to circumvent the problem. The run.sh is a wrapper script that calls the make command which calls the update_script.sh which updates the run.sh script - causing the issue. Since I cant do anything outside the run.sh command the only solution I would see is to "capture" the update_repo command in the run.sh script, and wait basically to the last second (before the scripts exits) with updating run.sh. Not even sure if that would work.

Brain teaser!

[gouttegd]

Not saying this is the right solution, but a separate launcher like the one evoked here and here would be one way to get rid of this issue, because the launcher would no longer be part of the files that are updated by update_repo.

I have a PoC, experimental binary launcher for the ODK in the works.

[matentzn]

WOW! Awesome! Surely keeping track of that!

[gouttegd]

Don’t expect fast progress! This is a side-project of a side-project.