Respect package.readme in Cargo.toml #1216
Conversation
Unfortunately, because #1185 took a dependency on a new crate, the problem needs to fix there as well. It has the same problem as this crate did where it expects a readme to be a path only (present or absent) so I can't tell the difference between explicitly disabled or just not specified. I've opened oli-obk/cargo_metadata#234 to track. |
Fixes rustwasm#1215 by honoring how package.readme is authored in Cargo.toml, if authored. This could be a path or a boolean to disable searching for the path. Also fixes the existing `it_copies_a_readme_provided_path` test which was an exact copy - just with fewer empty lines - of `it_copies_a_readme_default_path` and wasn't honoring any authored path since there was no way to get the authored path.
|
Ready. I misunderstood that cargo_metadata was consuming from the command |
|
Looks like the test failures started in or around 3073db8 |
Great, thanks a lot! Appreceite you fixing it!
Yeah, I will look into the failures. Thanks |
|
Sorry @heaths. I had to revert this PR since we got some actual test failures after this was merged. Both locally and in Github actions. Could you perhaps take a look at that and submit a new PR? Sorry for the inconvenience... |
|
Understood. Will take a look. |
|
To note, the two failing log_level tests started earlier in 3073db8 - sometime around when the deserialization of the manifest was removed from this crate in lieu of running a tool. |
|
Thank you for taking the time to investigate and fix it! Appreciate it! |
Fixes #1215 by honoring how package.readme is authored in Cargo.toml, if authored. This could be a path or a boolean to disable searching for the path.
Also fixes the existing
it_copies_a_readme_provided_pathtest which was an exact copy - just with fewer empty lines - ofit_copies_a_readme_default_pathand wasn't honoring any authored path since there was no way to get the authored path.