Le 19/07/2022 à 15:00, David Kastrup a écrit :
Jean Abou Samra <j...@abou-samra.fr> writes:
Meanwhile, like Lukas, I have been unashamedly switching
back to 2.22 when I wished a better diagnostic. This is
what we advise most users to use, so I'm not bothered.
The issue is marked as a regression, so it will have to
be fixed before we release the next stable version anyway.
What was the news regarding this issue on the current stable
(!#$#%!$%$#%!$) 3.0 release of Guile? What were the roadblockers there?
Anybody who has had enough involvement here to give a quick refresher?
Guile 3.0 still gives poor diagnostics (without error locations)
on evaluated code. Its latest release also gives poor diagnostics
(with <unknown location>) while byte-compiling, but this is not
problematic for the issue at hand. When I last looked, it
seemed that we'd have to use different APIs in Guile 2 and Guile 3
to get better diagnostics, but it's been a while, I don't recall
the details right now, and for personal reasons I am likely to
be less active for a while.
