Garbage like "Ö0Ï" in autocomplete suggestions for C library functions
The function signatures from autocompletion suggestions for the C standard library functions contain weird characters like Ö0Ï. For example, typing fputs suggests fputsÌ1024Í(const char * __s, FILE * __stream)Ö0Ïint.
Geany 2.0 from the standard package repository of Linux Mint 22 Wilma (Ubuntu 24.04).
Confirm with LM22 Geany 2.0 package.
Doesn't happen with Geany tag 2.0.0 compiled locally.
The standard loaded tags file std99.c.tags with the local build starts the line in hex for fgets with:
66 65 64 74 73 CC 31 30 32 34
f g e t s ?? 1 0 2 4
whereas the LM std99.c.tags (/usr/share/geany/tags/std99.c.tags) starts the line in hex for fgets with:
66 65 64 74 73 C3 8C 31 30 32 34
f g e t s ?? ?? 1 0 2 4
which is clearly the wrong format with C3 8C not CC after the name
Amazingly Geany still manages to load the tags, but of course is confused about the contents since its not the right format.
AFAICT LM just copies the package from Ubuntu so its the generated Ubuntu package that is wrong. Someone with Ubuntu needs to verify the package has a different std99.c.tags file and why? @hyperair maybe you can help?
Maybe it's because the maintainers of the Debian package convert the encoding?
https://git.launchpad.net/ubuntu/+source/geany/tree/debian/rules?h=ubuntu/noble#n33 https://salsa.debian.org/geany-team/geany/-/commit/1f2bd775333f01cf3e2d521f2c817ad8ee0247d6
Ok, yeah, the tags file is not a text format, it has entry lines with fields separated by binary bytes (eg the CC byte) so iconv cannot simply convert to UTF-8. The package building needs to convert each individual field, not treat a whole binary separated multi-field entry as text.
As you found it feel free to report it to Ubuntu as your credit.
Reported to Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1086558