[wip] Attempt to fix build #1769
Conversation
|
Hmm, @samphippen (given you're way more active with rspec-rails than I) what do you want to do about the cucumber segfaults, we can't update cucumber, we're on the last version supporting 1.8.7 anyway |
|
@JonRowe if this is still relevant I can take a look at it. Is this still a thing? |
|
We still need to fix the 1.8.7 build so yes? I've added a commit which skips cucumber on 1.8.7 which should help. |
1.8.7 is now segfaulting reliably for cucumber, and our smoke tests so we skip
|
A wild green Rails build appears! This is just skipping the acceptance on 1.8.7 though, @samphippen and @myronmarston how do you feel about this? I think given that 1.8.7 is so EOL and this only affects Rails 3 I'm ok with it, we're still running unit specs and confirming cross compatibility with the other gems and its unlikely any 1.8.7 specific errors will crop up here? |
The problem is that the unit spec coverage in rspec-rails is pretty poor, from what I remember (although I haven't worked much in this code base), and the cukes do a much better job of coverage. Also, 1.8.7 specific errors are very easy to do accidentally, e.g. wrong hash syntax or whatever. Couldn't we just add a |
|
No the issue here is the segfaults, the gem versions haven't changed. |
But what is different in the environment from when 1.8.7 builds passed? |
|
Nothing as far as I know |
| bin/rake acceptance --trace | ||
| return $? | ||
| else | ||
| return 0 |
There was a problem hiding this comment.
Might be worth echoing a warning explaining that this part of the build is being skipped and why.
|
OK. Ideally, it would be nice to just skip the part of the acceptance cukes that trigger this instead of all of them, but I'll defer to you and @samphippen on this. |
|
I don't think its that predictable, I'm happy for this to be a temporary measure but I don't have the time to investigate any further and it'd be nice to get them as green as possible. |
👍 |
1.8.7 is now segfaulting reliably for cucumber, and our smoke tests so we skip
No description provided.