Yes, I can backport the 7.x fixes to the 6.x release branch upstream. The result is a loss of the ability to set or retain PDF metadata fields other than pure ASCII. All characters that can't be represented in ASCII are getting replaced with '?'.
(As far as I can tell, the Ghostscript commit intended to removing Unicode metadata in DOCINFO for only PDF/A-1 but removed it for PDF/A-2 and A-3 as well, and this was likely not what they intended. I will report that separately in their bug tracker. It doesn't need to be dealt with in Debian.) On Sun, 16 Sep 2018 at 12:16 Sean Whitton <spwhit...@spwhitton.name> wrote: > Hello, > > On Sun 16 Sep 2018 at 08:40PM +0200, Paul Gevers wrote: > > > Dear Sean, > > > > On 16-09-18 20:30, Sean Whitton wrote: > >> Paul: does preventing regressions in testing take precedence? > > > > Normally, yes temporarily (we are not blocking yet), but ghostscript was > > uploaded with urgency high. That means that regressions are ignored and > > without an RC bug, ghostscript will migrate to testing tomorrow (if my > > counting is correct). > > Thanks. I should have been clearer that I was asking about /all/ > regressions in testing, rather than just autopkgtests. > > >> If so, > >> this bug should be reassigned to ghostscript and raised to RC severity. > > > > I don't follow what you mean by this sentence. > > I meant that if preventing a test suite failure in testing took > priority, we would want to stop ghostscript from migrating. Anyway, > it's moot. > > > ocrmypdf can stay in testing as long as it doesn't have an RC bug itself. > > > > So just to make it clear: if this change making ocrmypdf totally > > unusable and you still want ghostscript to migrate to testing to fix > > multiple CVE's, than assigning this bug to ocrmypdf and raising it to RC > > level will start the autoremoval process. If you think it is worth > > searching for a solution in ghostscript to avoid it breaking ocrmypdf, > > than reassign this bug to ghostscript and raise the severity to RC level > > to avoid migration. > > Right. > > I don't know if OCRmyPDF in testing counts as RC-buggy right now, > because I don't know the ramifications of the GhostScript changes; I'm > going to wait on upstream. > > -- > Sean Whitton >
