Resolver error formatting #3862
Conversation
segiddins
force-pushed
the
seg-resolver-error-formatting
branch
from
8c9292c to
0163e21
Compare
|
We should remember to apply the following diff once CocoaPods/Molinillo#24 is merged and released: diff --git a/lib/bundler/resolver.rb b/lib/bundler/resolver.rb
index f21c9cf..ec3b2a4 100644
--- a/lib/bundler/resolver.rb
+++ b/lib/bundler/resolver.rb
@@ -35,6 +35,9 @@ module Bundler
depth = 2
tree.each do |req|
t << " " * depth << %(#{clean_req req})
+ if spec = conflict.activated_by_name[req.name]
+ t << %(, resolved to #{spec.version},)
+ end
t << %( depends on) unless tree.last == req
t << %(\n)
depth += 1 |
|
@indirect this is now ready for review |
segiddins
force-pushed
the
seg-resolver-error-formatting
branch
from
c9a82ec to
8be3be3
Compare
| rack_middleware (>= 0) ruby depends on | ||
| rack (= 0.9.1) ruby | ||
| rack-obama (= 2.0), resolved to 2.0, depends on | ||
| rack (= 1.2) |
There was a problem hiding this comment.
Now that I'm reading it, I'd prefer rack-obama (= 2.0) was resolved to 2.0, which depends on\n rack(= 1.2). Sorry to nitpick on you. 😝
|
This is a huuuuuge improvement in the UX of requirement conflicts. 👏 🙌 |
|
@indirect done |
|
Woo! @homu r+ |
|
📌 Commit 210f9d9 has been approved by |
Resolver error formatting Closes #3803
|
💔 Test failed - status |
|
@segiddins looks like it's broken on 1.8.7 |
|
Looks like something has an unstable comparable implementation to me? |
|
Hashes literally do not have a stable iteration order in 1.8.7; you'll need to sort the hash before printing the conflict message or store conflicts in an array of tuples. |
|
@indirect should work now! |
|
@homu r+ |
Resolver error formatting Closes #3803
|
💔 Test failed - status |
segiddins
force-pushed
the
seg-resolver-error-formatting
branch
from
67b0274 to
f09b2c6
Compare
segiddins
force-pushed
the
seg-resolver-error-formatting
branch
from
f09b2c6 to
ef5b0ea
Compare
|
@homu r+ |
|
📌 Commit ef5b0ea has been approved by |
Resolver error formatting Closes #3803
|
💔 Test failed - status |
| if conflict.locked_requirement | ||
| o << %( In snapshot (#{Bundler.default_lockfile.basename}):\n) | ||
| o << %( #{clean_req conflict.locked_requirement}\n) | ||
| o << %( #{conflict.locked_requirement}\n) |
There was a problem hiding this comment.
This is actually a Gem::Dependency, whose to_s method includes "runtime" on RubyGems 1.5 and older. How is it not a DepProxy? No one knows...
|
@homu r+ |
|
📌 Commit 3fad388 has been approved by |
Resolver error formatting Closes #3803
|
☀️ Test successful - status |
|
Wooooo! Finally! Thanks for the help on this one, @indirect ! |
|
ugh, just realized that I should be reading the platform from the dep... the output will be wrong if the platform isn't Ruby. :P |
|
mhmm :P |
|
Actually, @indirect, do we even have the platform for locked gems? |
|
no idea! that's... kind of distressing, actually. |
|
Opened #3906 to print platform(s) correctly. |
Closes #3803