Changed appveyor.yml to reduce tests #579
Conversation
|
Hmm.. One thing to consider here is the order. I wonder if there's any way to get statistics on which environments fail most frequently (preferably where builds where all platforms failed and builds that did not finish are discarded). My instinct is to order this like this:
But I think that's probably not the right order. Probably the right order (in terms of failing fastest) is this:
So.. basically exactly what you did, except putting the Let me know if you feel strongly about this, this is just kinda shooting from the hip. |
Codecov Report
@@ Coverage Diff @@
## master #579 +/- ##
=======================================
Coverage 94.87% 94.87%
=======================================
Files 28 28
Lines 7184 7184
=======================================
Hits 6816 6816
Misses 368 368Continue to review full report at Codecov.
|
|
I'm not sure if there's a way to get statistics, I'll do some research on it. For now I'll go with the order you mentioned second (x86 at the end). |
|
LGTM if @gaborbernat agrees. |
appveyor.yml
Outdated
| platform: | ||
| - x86 | ||
| - x64 | ||
| - PLATFORM : x64 |
There was a problem hiding this comment.
technically correct, but a shorter way of expressing this (and smaller change) is just banning the three excluded envs (https://www.appveyor.com/docs/build-configuration/#exclude-configuration-from-the-matrix)
There was a problem hiding this comment.
I believe this was @kirit93's original approach. I'm not sure why he abandoned it.
There was a problem hiding this comment.
@pganssle, @gaborbernat I thought it would be simpler to play around with the order of the jobs if I explicitly defined the matrix. If you take a look at the latest commit that uses exclude, the jobs are not ordered in the way we want them to be.
Since we want all the x64 jobs done first followed by the two x86 jobs, the previous commit has the right implementation.
|
Hm.. Appveyor does not seem to be clearing the queue, I'm not sure why. I think it's unrelated to this change. |
Related to issue #529
@gaborbernat Please let me know if the changes I've made are what is expected.