control: tags -1 + important Le mardi 29 octobre 2024, 17:18:03 UTC gregor herrmann a écrit : > On Tue, 29 Oct 2024 16:08:30 +0000, Niko Tyni wrote: > > > This gives a list of thirteen integers on trixie, but > > just one undef on sid. > > > > Is this an intentional API change in ImageMagick 7 that GD::SecurityImage > > should adapt to, or possibly a regression on the ImageMagick side? > > Nice finding. > > I also have a small info point to add. > > The first of the 2 (now) failing test looks like this (same for the > second in principle): > > #v+ > # cat debian/tests/perlmagick-gd-7.q16 > #!/usr/bin/perl > # regression test for #1032998 > use GD::SecurityImage use_magick => 1; > > my $image = GD::SecurityImage->new( > width => 200, > height => 100, > lines => 4, > gd_font => 'Giant', > scramble => 1, > rndmax => 10, > ); > $image->random; > $image->create( 'normal', 'default', "#403030", "#FF644B" ); > print $image->out( force => 'png' ); > #v- > > Now GD::SecurityImage has two backends: GD::SecurityImage:GD has a > gd_font property but GD::SecurityImage::Magick only has a font property > (which needs an absolute path to a *.ttf file). And you are of course > using the ::Magick backend :) (with use_magick => 1). > > I wonder how this has ever worked; with use_magick gd_font attribute > is never used (because it's in GD.pm) -- which amounts to Niko's > finding where he used an empty 'font'. > > > If I change your test to > > #v+ > # cat debian/tests/perlmagick-gd-7.q16 > #!/usr/bin/perl > # regression test for #1032998 > use GD::SecurityImage use_magick => 1; > > my $image = GD::SecurityImage->new( > width => 200, > height => 100, > lines => 4, > gd_font => 'Giant', > font => '/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf', > scramble => 1, > rndmax => 10, > ); > $image->random; > $image->create( 'normal', 'default', "#403030", "#FF644B" ); > print $image->out( force => 'png' ); > #v- > > i.e. I add a 'font' value (and I picked DejaVuSans.ttf randomly as it > was installed in the chroot where I built imagemagick; gd_font can be > removed), it passes. > > > > I suppose a depends on fonts is missing: > > Please elaborate. Did ImageMagick 7 change something in (default?) font > > handling? I couldn't easily find a summary of the changes. > > My tenatative conclusion so far is: > - The test was always incorrect, and passed by chance. > - This luck is gone as apparently something changed in imagemagick.
imagemagick is fixed and return always a default font on debian. So no more RC but a bug that need to be fixed in order to bail out early on undef. I suppose if you pass an incorrect font it will still do the same rouca > > > > Moreinformation here: > > > https://ci.debian.net/packages/i/imagemagick/testing/amd64/53694127/ > > Printing the generated image to the terminal at the end of your > > autopkgtest check seems useless and clutters the log BTW. > > Also my terminal :) > > > Cheers, > gregor > >
signature.asc
Description: This is a digitally signed message part.
