Skip to content
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

Enable Travis CI #42

Closed

Conversation

jakirkham
Copy link
Member

Fixes #10
Closes #34

Attempts to get macOS packages built by enabling Travis CI.

Drop the `skip` that was set for macOS builds. This was added only
because Travis CI was running out of time trying to build VTK. In the
interest of attempting to make macOS packages, remove the `skip` so we
can get Travis CI builds going.
Now that the skip for macOS is removed, re-render to enable Travis CI
builds. Hopefully we can now get some macOS packages.
@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

Copy link
Contributor

@Korijn Korijn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fingers crossed for build times...

Set the number of cores that ninja can use explicitly to make sure it is
being set correctly and it is not overestimating or underestimating.
This is really not a good idea and no one should do it. Still sometimes
we seem to be able to get a little more from Travis CI if up the degree
of parallelism used.
@Korijn
Copy link
Contributor

Korijn commented Sep 22, 2017

This is a long shot, but the most recent version of ninja (v1.8.2) has the following line in its changelog:

Use POSIX_SPAWN_USEVFORK with glibc, makes some full builds twice as fast #1182

Since we are using ninja v1.7.2, bumping it might do the trick? For this purpose, I opened conda-forge/ninja-feedstock#5


On a different note, I read in the Kitware blog that ninja defaults to building with cpu_cores + 2 jobs.

@Korijn
Copy link
Contributor

Korijn commented Sep 22, 2017

Now that ninja v1.8.2 is out, we can just retrigger the build to see if it makes any difference. I've restarted the Travis CI build.

I would recommend switching back to just ninja install and letting ninja determine the amount of parallelism.

@Korijn
Copy link
Contributor

Korijn commented Sep 22, 2017

Still getting timeouts...

More recent versions of clang (that come with more recent versions of xcode) are also known to have faster compile times. Why is conda-forge on osx_image: xcode6.4? It seems osx_image: xcode9 is much more recent, and might provide faster build times through newer versions of clang.

@jakirkham
Copy link
Member Author

Why is conda-forge on osx_image: xcode6.4?

The reasoning is similar to why we use CentOS 6 for Linux builds. We try to build against the oldest OS that we want to support. That way we can guarantee that everything should work on that OS or any newer version. This xcode6.4 has macOS 10.10, but includes the SDK for macOS 10.9. As conda-forge started out supporting macOS 10.9 and we haven't discussed changing the lowest version of the OS, we continue to try to do so. Admittedly at some point this will have to change this.

...might provide faster build times through newer versions of clang.

That's certainly something we can explore. Though the macOS version we target and the clang version used can be varied independently. We build clang in conda-forge in a package called clangdev. So we could try using that. We would need to require libcxx as well for C++ support.

@Korijn
Copy link
Contributor

Korijn commented Oct 17, 2017

I'm closing this as it doesn't look like VTK will ever build within the Travis CI build time limit constraints.

@Korijn Korijn closed this Oct 17, 2017
@Korijn Korijn mentioned this pull request Oct 17, 2017
@jakirkham jakirkham deleted the enable_travis_ci branch May 12, 2018 21:18
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants