> Because the backtrace actually doesn't even point in the right place, it's
> quite useless in my opinion.
the worst part of it is that it is able to tell you the right file and line,
but it doesn't tell you the name of the function.
> Am I doing anything wrong? I use guile 3.0.1.
no (although there are newer versions).
i believe that most of the complaints regarding guix error reporting would be
gone if guile had usable backtraces.
pardon my bluntness, but it still baffles me how can anyone work on guile
longer than a day without dropping everything in frustration, and shift focus
on fixing its backtraces. in other contexts i did that for way less.
heck, even i once ventured there already, in my early days with guix! (but then
quickly gave up when i realized that most of guile is written in C)
sidenote: that is when i also asked why guix is not scheme agnostic and makes
it possible to run on top of another scheme with better error reporting...
don't ask! since then i have recalibrated my false assumptions about
portability in the scheme ecosystem.
meta: the reason i'm complaining is not because this is the state of things.
i'm even willing to help with that! but rather because i don't see it
acknowledged that this is way less than ideal. including that too much of guile
is written in C, and that the contact surface between guix and guile is way
bigger than necessary.
--
• attila lendvai
• PGP: 963F 5D5F 45C7 DFCD 0A39
--
“Good people don’t need laws to tell them to act responsibly, and bad people
will find a way around the laws.”
— Plato (c. 427–347 BC)
