On Mon, 10 May 2021 at 20:30:42 +0200, Guilhem Moulin wrote:
> On Sun, 09 May 2021 at 17:48:28 +0200, Guilhem Moulin wrote:
>> That said I'm not sure to how to fix this.  I'm not really familiar with
>> lighttpd but I don't see a way to list enabled modules other than
>> looking in /etc/lighttpd/conf-enabled which I'm not really keen to do.
>> […]
>> lighttpd maintainers (CC'ed): What's the best way to express “enable
>> fastcgi-php (or fastcgi-php-fpm) unless fastcgi.server already has a
>> handler for .php”?  Or if that's not possible “enable fastcgi-php unless
>> fastcgi-php-fpm is already enabled”?  Simply expand
>> /etc/lighttpd/conf-enabled/*-fastcgi-php-fpm.conf and check for a match,
>> or is there a more robust way?
> 
> Here is a workaround that seems to work:
> 
> https://salsa.debian.org/roundcube-team/roundcube/-/commit/4638e975366c980280bd3b63020eaee45df88f52
> 
> lighttpd maintainers: is it acceptable to look for matches in
> /etc/lighttpd/conf-enabled/*-fastcgi-php.conf and …/*-fastcgi-php-*.conf?
> The aim would be to ship a fix in Bullseye; we can always refine the
> logic during the Bookworm release cycle.

FWIW I plan to upload this some time next week and ask the Release Team
for an unblock request.

Given #898040, #988236 and #988282 I believe the vast majority of
roundcube-core users don't use our lighttpd integration.  Nonetheless I
added an autopkgtest (in experimental) to help detect such regressions
further down the road. :-)

Cheers
-- 
Guilhem.

Attachment: signature.asc
Description: PGP signature

Reply via email to