Closed Zweihander-Main closed 6 months ago
Also, have a look at the CI failures. In general, this PR looks good.
I've added (eval-and-compile (require 'lisp-mnt))
for the CI failure related to byte compilation and updated/rebased the branch to avoid copyright year errors from the eldev doctor
part of the pipeline -- think that should pass now.
Thank you for the fix!
Ran into the following bug using the new source directories feature and
eldev info
:README-elpa
,README-elpa.md
,README
,README.rst
orREADME.org
file in the root of the project. See here for why those particular files.eldev -d info
.Result (Using Eldev's
test/dependency-d
directory as an example):This PR fixes that but two issues I want to flag --
package--get-description
first checks for README files and then if they aren't found, useslm-commentary
. Rather than attempt to advise/rewritepackage--get-description
with the source directory, this catches it if it errors with the assumption that the README wasn't found and the error was in thelm-commentary
half of it. It then runslm-commentary
on the correct file (and doesn't bother trying to find a README in a source directory). The possible problem here is it won't catch any regressionspackage--get-description
may turn up in the future -- it always assumes an error there is going to be this one source directory case.eldev-info-missing-dependency-1
test suggests that different versions of Emacs may turn up the commentary string or the package title. The second source directory test is trying to confirm that if the commentary is present in the main code file, it'll come up but since it has to test for both, it doesn't actually make any difference if you just catch thepackage--get-description
error and then use the source directory to find the commentary string or if you just use(error nil)
after catching it and letting the next block of code handle it.If this is the wrong approach here, happy to try again or otherwise feel free to discard; mostly wanted to flag the problem!