On Thu, Nov 20, 2025 at 10:53:40PM +0100, Matthias Geiger wrote: > > /usr/bin/osx:./linux-gpib.sgml:426:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:429:5:E: element "ROW" undefined > > /usr/bin/osx:./linux-gpib.sgml:430:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:431:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:437:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:438:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:440:5:E: element "ROW" undefined > > /usr/bin/osx:./linux-gpib.sgml:441:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:442:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:446:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:447:7:E: element "ENTRY" undefined > > /usr/bin/osx:./linux-gpib.sgml:449:5:E: element "ROW" undefined > > > I suspect something in the build-depends did change, as -2 and -1 built > fine. I can repro this locally; strangely, nothing in /usr/bin/osx changed, > so I'll have to check where the build fell apart.
In case it helps, I did a debbisect and this is the outcome: bisection finished successfully last good timestamp: 20251025T204106Z first bad timestamp: 20251026T082559Z the following packages differ between the last good and first bad timestamp: apt 3.1.9 -> 3.1.10 libapt-pkg7.0:amd64 3.1.9 -> 3.1.10 Then I compared the "last good" build log with the "first bad" build log and found this: --> The "docbook" package is installed in the "last good" but it's not in the "first bad". So I see that one of the two following things is probably true and the other is false: A) this was a bug in the old apt, the new apt is actually right in not installing "docbook", and this is also a bug in your package, which should build-depend on "docbook". B) this is only a bug in the new apt, which should install "docbook" but it does not. I'm curious about what of those two is the truth... Thanks.
