Email or username:

Password:

Forgot your password?
Top-level
Q

@jwildeboer I've been contributing/following a few big projects that only use mailing list based contributions (Yocto, U-Boot bootloader, Linux kernel, libcamera, Buildroot). The last few years we've seen a lot of people voicing their discontent at that workflow and requesting we do everything with GitHub/GitLab "or else you will never get a contribution from me".

So I am not sure those young people are that interested in that old workflow.

14 comments
Jan Wildeboer 😷:krulorange:

@0leil And I support them with that. The very old workflow of creating patch/diff and sending that around is dark magic :) With issue and PR management a la Github/Gitlab/Gitea/Codeberg a lot has become a lot better.

Q

@jwildeboer well... it is **a** workflow... which has nothing in common to mailing list based contribution.

GitHub/GitLab's PR workflow is an absolute disaster/nightmare for per-commit/patch review. Gerrit does that a bit better with patchsets but it gets difficult to read pretty quickly, even with topics.

GitLab/GitHub is just NOT necessarily compatible with some project's workflow.

I actually wrote a multiple pages long email on why no GH/GL for Yocto a while back, maybe I should put it in an article so it's easier to explain everything instead of being limited to a few 100s of characters at once :)

@jwildeboer well... it is **a** workflow... which has nothing in common to mailing list based contribution.

GitHub/GitLab's PR workflow is an absolute disaster/nightmare for per-commit/patch review. Gerrit does that a bit better with patchsets but it gets difficult to read pretty quickly, even with topics.

Jan Wildeboer 😷:krulorange:

@0leil Please do! I will also start a blogpost on how Discord etc mess up communication in a project (IMHO) by mixing synchronous and asynchronous communication together in a way that isn't helpful but, again IMHO, damaging to the flow.

Anthk

@jwildeboer @0leil It sucks, period. A mail list/usenet gives you all anytime.
And IRC will work even with a 386 in a hurry.

Jonathan Frederickson

@0leil @jwildeboer Per-commit review is a point that I hadn't considered. With GitHub's PR UI only making the end result of all commits together apparent, I could see someone introducing a security vulnerability in a commit, fixing it in another, and then tricking a packager into publishing the version with the vuln...

ManicDee

@jfred @0leil @jwildeboer per-commit doesn’t make sense when you are just going to deliver all those commits in one PR, same effect as squashing commits when accepting the PR.

Main issue I see is trying to do too much in one PR.

Q

@jwildeboer ah and before I forget, there's this tool called b4 now that aims at making patch sending for mailing lists a bit easier. I personally have switched to that for the last 1+ year, I am NOT looking back. git-send-email? I now say no no :)

Q

@jwildeboer

I'm still in my early thirties but I think I was probably one of the last generations where GitHub was not in basic monopoly for anything-git. At university it was still a decision which to use and some were still using Dropbox :)
So I think it was less of an issue back then, because not EVERYTHING was on GitHub. Now it gets more difficult to get people to do a bit more effort to contribute, is my feeling.

Jan Wildeboer 😷:krulorange:

@0leil I come from the time when version control was an esoteric topic at first, tarballs and patches ruled. Then I started using CVS and along came Sourceforge, which many projects used. That was what git and GitHub are now, more or less. So yes, better tools, but still the same (centralised) setup. Forgejo/Codeberg are working on integrating ActivityPub by using #ForgeFed [1] and I have high hopes that that will introduce truly better ways and real progress.

[1] forgefed.org

loleh πŸ’Ύ

@jwildeboer @0leil interesting! Does this mean we might be able to follow a @repository@codeberg.org one day soon?

EmpathicQubit/Earth 616 Ver.

@0leil @jwildeboer I'm around the same age, and in university for my group's final project I literally created a user on my computer for Git and people SSHed into it

unusual zone of infecundity
@0leil @jwildeboer i wonder how much the learned helplessness that comes from overreliance on these monopolizing monolith platforms contributes to people's later inability or unwillingness to contribute
Kevin Granade

@0leil @jwildeboer after running a largeish project for a decade now I am immediately skeptical of anyone saying, "you won't get contributions from me unless you change something".

In practice this has always landed somewhere between "when pigs fly" and "don't threaten me with a good time".

Q

@kevingranade @jwildeboer The issue is that most FOSS projects are underfunded, understaffed, overworked or its devs/maintainers on the brink of burn out or - most likely - all at the same time. Having more contributors and especially retaining them is crucial, so it's not always easy to say "please, more contributors" and then shun people away with "yes but not by changing our contributor workflow". If we keep hearing from different people we lost potential contributions because the contribution workflow was unacceptable to them, we have to decide whether the loss of their theoretical contribution was worth it. I don't have the answer.

@kevingranade @jwildeboer The issue is that most FOSS projects are underfunded, understaffed, overworked or its devs/maintainers on the brink of burn out or - most likely - all at the same time. Having more contributors and especially retaining them is crucial, so it's not always easy to say "please, more contributors" and then shun people away with "yes but not by changing our contributor workflow". If we keep hearing from different people we lost potential contributions because the contribution...

Go Up