Le 17/03/2024 à 21:57, Harald Anlauf a écrit :
Hi Mikael,
thanks for the patch!
Regarding the first part of the patch, I think that fixing bad testcases
can be done at any time. Retaining identified, broken testcases means
that one may hit bogus regressions, hindering progress.
The second
Hi Mikael,
thanks for the patch!
Regarding the first part of the patch, I think that fixing bad testcases
can be done at any time. Retaining identified, broken testcases means
that one may hit bogus regressions, hindering progress.
The second part of the patch looks at first glance fine to
This fixes a spurious invalid variable in specification expression error.
The problem is caused by improper restoration of formal_arg_flag to false
(instead of restoring it to its previous value). This happens with the
testcase from the PR where a dummy argument is itself a procedure with dummy