Tags: confirmed
It also reproduce with other non-UTF-8 locales.
1. Edit "/etc/locale.gen" and setup non-UTF-8 locale
2. Use non-UTF-8 locale for Calibre
$ LANG=ja_JP.EUC-JP calibre
$ LANG=en_US.ISO-8859-15 calibre
3. Segmentation fault
> If LC_CTYPE=ro_RO then calibre segfaults on startup
> If LC_CTYPE is unset (or LC_ALL=C) then calibre starts and works normally.
> Locale: LANG=en_GB.UTF-8, LC_CTYPE=ro_RO (charmap=ISO-8859-2),
> LANGUAGE=en_GB:en
Current Linux system recommends UTF-8 based locale.
"ro_RO" uses legacy encoding "ISO-8859-2" and it's not recommended for
modern Linux systems.
Use UTF-8 based ro_RO locale "ro_RO.UTF-8" to avoids this bug.
1. Edit "/etc/locale.gen" and adds "ro_RO.UTF-8" locale support.
Use "dpkg-reconfigure locales" command for easy setup.
It also drops legacy locale support.
# dpkg-reconfigure locales
2. Use "ro_RO.UTF-8" for Calibre
$ LC_CTYPE=ro_RO.UTF-8 calibre
$ LANG=ro_RO.UTF-8 calibre
3. Works well.
> The offcial calibre version 5.43.0 does not segfault on the same machine.
> This is a debian specific issue.
Official calibre uses embedded Python, and not uses system Python.
Embedded Python uses limited locale support, and this avoids locale problems.
--
YOKOTA Hiroshi
