-
-
Notifications
You must be signed in to change notification settings - Fork 229
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
Improve travis-ci.sh #1833
Improve travis-ci.sh #1833
Conversation
Thanks for your pull request, @Geod24! |
Put the 'regular' test code at the beginning, and the part that relies on COVERAGE being defined only once at the end. Those could eventually be split in the future.
3ce4866
to
2134119
Compare
Ugh this stage setup is also terrible. If something fails in the first stage and is retriggered (and pass), the second stage never starts so the whole build is marked as 'cancelled' and one has to restart everything. Also the reason why #1795 happens is pretty clear, looking at |
Yeah it was on my list already. I will be travelling a lot for the next few days, but I think I can get this up before New Year's Eve. |
BTW the reason for running the second stage for every PR was that once we enabled the Travis builds which upload to GitHub we weren't sure how reliable they are and it was also the only way to test the PRs. However, a solution would be to enable them only for a PR which is from the dub repository and not with branches from forks. |
I don't quite get why we do that ? IIUC, the original intent was for this stage to trigger only on tags: Lines 105 to 107 in 2e9a26a
Which makes much more sense to me. I couldn't find much in the original PR. I guess whatever bug you were working around at the time might have been worked out, since LTO has seen a lot of love over the last 2 years ? |
So that the people can download dub binaries (from GitHub releases). This is interesting for example if you want to have a newer version of dub with an older compiler.
Yes, though as I mentioned this wasn't done as otherwise the scripts couldn't be tested. I fully agree that there should be no need to run them for a normal PR build. I was just explaining the historical reason ... or at least trying to.
Yes. |
Okay, I think revisiting the travis config can be done later though. This is good to go as is and already gave me useful insight. |
This will probably break horribly, but there's only one way to find out...