Fixes CI configuration and make tests passes #260
Conversation
|
@jcupitt sorry, have not verified linter CI before the merge. Please check here, we now it should works |
|
some tests are failing will fix them soon |
|
No problem, and thanks again for working on this. |
11bbc6f to
a7942cb
Compare
|
@jcupitt when I use libvips 8.9.1 there is warning like I thought this is only truffleruby problem only, but now I see that for ubuntu 20, libvips does not update the field for ruby 2.7.2 as well. Do you know why it could be? |
|
Yes, there was a breaking change in the way metadata modification was handled in libvips 8.9, unfortunately :( 8.10 handles this more gracefully. I should add something to the ruby-vips test suite to skip these tests on that version. |
|
Off-topic, but I suppose rubocop can be removed now that you've switched it to standardrb, is that right? |
|
@jcupitt so if I skip ti for 8.9 version that is expected behavior? |
|
@jcupitt I'll remove rubocop in spearate pr, that will be epic PR with all files changes |
|
Yes, sorry. I'll make a PR that fixes the test suite with 8.9. |
|
(this is partly why the old travis CI script built a specific version of libvips ... it reduced the complexity from having ruby-vips and libvips versioning decoupled) |
|
Cool fixed for ubuntu 20, found that 18 and 16 has some broken test, will handle them in different pr |
|
@jcupitt please check and merge, all tests should be green. But I added some version verifications |
| RSpec.describe Vips::Source do | ||
| it 'can create a source from a descriptor' do | ||
| source = Vips::Source.new_from_descriptor(0) | ||
| RSpec.describe Vips::Source, version: [8, 9] do |
There was a problem hiding this comment.
Oh nice, I didn't know about that.
|
Nice, lots of useful fixups. |
No description provided.