-
-
Notifications
You must be signed in to change notification settings - Fork 2k
Fixed missing constant with rubygems integration #4981
Conversation
failing result https://travis-ci.org/bundler/bundler/jobs/159755964 seems not related this change. Can you retry it? |
Restarted the failing build. |
@indirect ❤️ |
Is there any way to add a test? What's the constant being referenced? |
It's hard way to add test. Because bundler has no test with
It's |
The reason of this problem is I sent workaround Pull Request with explanation before this, but it's bad. Because it doesn't resolve the problem of Why does Bundler use |
Why isn't the autoload working here, @asutoshpalai ? |
I added workaround to ruby/rake :( |
@indirect @segiddins I understood your concerns. I try to code minimum reproduction case. |
Sorry for rough presentation... this problem is difficult to solid discussion. require 'bundler/gem_tasks'
require 'rdoc'
task default: nil This |
@indirect Yes, it is. I made sure of it on my local. But I think that it may be still workaround, because I don't understand what is real problem. |
@indirect I created a branch that uses
|
@asutoshpalai ping? |
I have been trying to reproduce this locally for some time with no success. But from what I can infer (and I believe is known to everyone in the discussion) from the proposed fixes, I can say that autoloading When As for the fixes, I think we should go with aycabta/bundler@4c32972d . But it's elusive reproduction (and lack of test case) makes me think that the problem is a bit deeper and may recur. |
I tentatively created Pull Request with aycabta@4c32972d: #5010. |
Workaround a bundler issue which is tracked on rubygems/bundler#4981
…gin-api, r=indirect [workaround] Use `require` instead of `autoload` for bundler/plugin/api Please read #4981 (comment). > But it's elusive reproduction (and lack of test case) makes me think that the problem is a bit deeper and may recur. I think so. This Pull Request is just *workaround*.
@asutoshpalai said:
and I created #5010 with aycabta/bundler@4c32972. For the last time, it was merged. I think this problem is not clearly solved. #5010 is just workaround. But this problem will not presumably occur again because |
I will confirm reproduction of this issue after bundler-1.13.3 release contained #5010 |
I confirmed to fix this with bundler-1.13.3. ref. ruby/rake#168 |
@hsbt bundler-1.13.3 was a single change release to deal with some rubygems.org server issues we experienced today. #5010 (and the rest of the planned 1.13.3 changes) is going to go out in bundler-1.13.4 Although if your issue is fixed, then that's great! |
…gin-api, r=indirect [workaround] Use `require` instead of `autoload` for bundler/plugin/api Please read #4981 (comment). > But it's elusive reproduction (and lack of test case) makes me think that the problem is a bit deeper and may recur. I think so. This Pull Request is just *workaround*.
@lynnco Thanks your information. Maybe my issue was solved by Travis environment. I reopen or submit new issue when I got same error. |
Fixes #5000.
I got uninitialized constant error with bundler-1.13.1 on travis. see https://travis-ci.org/ruby/rake/jobs/159734476
It's reproduce my local box with ruby/rake testing. I simply fixed this error.
Thank you.