-
-
Notifications
You must be signed in to change notification settings - Fork 32.2k
[3.3] Backport CI config from master #2481
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
Conversation
This PR is more or less a copy of the PR #2475 adapted for the 3.3 branch. Right now, I didn't backport the test_os change, since I got conflict and I'm not sure that the test_os os.walk() test also fails on 3.3. |
The Travis CI job failed with:
The AppVeyor job is full of errors, it seems like the build system is unable to open the Python 3.3 PCbuild files. |
@zooba, @zware, @brettcannon: Which Visual Studio is needed to build Python 3.3 on Windows? Any idea how to adapt AppVeyor configuration to fix it? |
If it's just impossible to build Python 3.3 on AppVeyor, hum, too bad. We might remove the AppVeyor CI in that case. If at least we could have Travis CI for Linux, it would already be a first enhancement! |
No idea, never tried to build it. I'd assume it's the same configuration as 3.4. Looking at the build log, it probably needs to be pointed specifically at the VS 2010 MSBuild.exe, rather than any newer version. |
I added 2 more commits:
test_protocol_sslv3() and test_options() of test_ssl still fail because of SSLv3. OpenSSL now disables SSLv3 by default: we should backport a9dcdab from the 3.4 branch, but I would prefer to do that in a separated commit since this change not only changes tests but also Modules/_ssl.c. |
Ok, I removed the [WIP] tag: I consider that this PR is now ready for a review. Summary of the PR for Python 3.3:
CI status:
|
I think getting AppVeyor working with 3.3 is probably more effort than it is worth. Various rewrites of the batch files in PCbuild (build.bat, rt.bat, etc.) were not backported that far, and probably don't need to be. FTR, 3.3 was built with VS 2010. |
I agree with not trying to get AppVeyor working. 3.3 is just about to be end-of-lifed. Basically, we just need to get the existing small backlog of security patches and PRs updated and merged and then that will likely be it for 3.3, assuming no new security issues arise between now and 2017-09-29. In any case, let's hold off on making a decision whether to merge this until we've had a chance to review the 3.3 backlog. |
.github/appveyor.yml
Outdated
@@ -0,0 +1,28 @@ | |||
version: 3.3.6+.{build} |
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.
As noted in the comments above, let's not try to deal with appveyor. There should only be a handful of further commits to the 3.3 branch before it is EOLed. So delete this file.
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.
I deleted AppVeyor.
.github/CONTRIBUTING.rst
Outdated
@@ -0,0 +1,47 @@ | |||
Contributing to Python |
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.
There's no reason to add this file for 3.3. Just delete it, please.
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.
deleted
Add .travis.yml for Travis CI.
The regen-all Makefile rule doesn't exist in Python 3.4, only since Python 3.5 and newer (and 2.7).
tzdata resource doesn't exist in Python 3.4.
Oh, Travis CI success on Linux: "All checks have passed" Strange. On my Fedora 25, test_ssl and test_ftplib fail (maybe an error related to SSLv3, I don't know exactly). Well, it's a good thing, the CI pass :-) |
The macOS CI has 4 failing tests: test__osx_support test_distutils test_socket test_sysconfig. Examples: @ned-deily: Should I try to fix these failures, remove the macOS CI from Travis config, or just do nothing?
|
Summary of the PR:
AppVeyor and CodeCov are not part of this PR, I consider that we don't need them for Python 3.3. For AppVeyor, it's just too hard to setup the CI and it was decided that it's not worth it (read comments above). IMHO a working CI on Linux is already a huge improvement compared to the current state: no more CI at all. We removed the Python 3.3 buildbots long time ago...
|
"The macOS CI has 4 failing tests." You should remove the macOS CI configuration. Vanilla 3.3.x has never functioned properly on macOS versions beyond OS X 9. 3.3.x was already in security-fix mode when OS X 10.10 was released which broke various things. It would be out-of-scope to try to fix things now as there has been no demand for it that I'm aware of and it would be sending the wrong message now; in any case, some downstream distributors have backported 3.3 support long ago if anyone really needs it. |
Ok, done. |
Yeah! A CI for 3.3, that's so cool ;-) |
No description provided.