-
Notifications
You must be signed in to change notification settings - Fork 18
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix sub-venv #145
Fix sub-venv #145
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Except for the venv deactivation thread, this is all resolved imho and can be merged.
703735d
to
1342c4f
Compare
CI passes.
This has been resolved. ("Because not all module selections depend on IM. For example, if you exclude ocrd_olena, Docker will probably not have ImageMagick. (That Dockerfile is not only used for pre-built configs!)")
Done
Done (see above) |
Signed-off-by: Stefan Weil <sw@weilnetz.de>
There's still a similar problem for wget |
Wow. This is yet another portability nightmare. We apparently cannot rely on Now what do we do? Crawl out and beg for |
Oh, got it: For one, |
- avoid .EXPORT_ALL_VARIABLES, because it forces expanding the error function unconditionally in the first recipe - instead, export PYTHON and VIRTUAL_ENV specifically - to ensure git/parallel can get installed along with deps-ubuntu, use recursive call for all deps-ubuntu-modules and CUSTOM_DEPS
Co-authored-by: Stefan Weil <sw@weilnetz.de>
Well,
How did you get a different result? If it works for you, too, I suggest to remove all hacks again. And there is no cache problem with |
@bertsky, I now merged most of your commits, especially the urgent bug fixes. Please rebase your PR, so we can discuss the remaining ones easier. |
You didn't say what your result was ;-)
But if I add a semicolon to the end of the
Anyway, we don't need any Meanwhile, I have also found the actual reason why our It's even possible to run |
What the ... |
@stweil you must be joking! You made me undo most of my PR after 3 days back and forth, and then just before I finally come around fixing the last issues, you just drag the commits out of the PR without even asking? I just reset master to the previous state, so we can wait for CI in the current, fixed state here and then merge if everyone is satisfied (as it should be). |
I had the impression that it was time to fix some issues after a whole week of discussion. So I applied those patches where we agreed that they are fine. What is the problem with this? Rebasing @bertsky's remaining commits is easy. I have already done in locally and could push that state to a new branch if that helps. |
We are all eager for a fix, but it should be based on consensus and successful CI. Cherry-picking individual commits and pushing them to master without consultation is just not good practice. We all have the power to push to master but we should only use it once we're in agreement. We were close but not yet there.
Again, it's not a question of technicality but of courtesy. And you are - rightly - an adherent of concise pull requests and static code analysis. Cherry-picking from an almost finished PR with 90+ discussion threads without asking first here or in gitter is not good. I appreciate that the pace of discussions can sometimes be frustrating but finding consensus improves the product in the long-term. |
@stweil Also, you forget that rebasing then becomes much harder and a one-way direction. In this concrete case, I had already rebased this PR locally, fixing the early commits regarding the sem and wget failures (which you yourself have been pointing out mostly). We were in the middle of a discussion about these. By dragging them onto master directly, you made it impossible to edit/undo/reorder them. After all, that's what a PR is for: aggregating a set of mutually supporting/consistent commits that have to be considered and developed together until mature enough to integrate at once. |
Fixes #140
Fixes #144