On Mon, 11 Nov 2024 at 09:55, Andreas Tille <ti...@debian.org> wrote:
> Am Thu, Nov 07, 2024 at 08:17:44PM +0100 schrieb Andreas Tille: > > Am Thu, Nov 07, 2024 at 08:34:47AM +0100 schrieb Petter Reinholdtsen: > > > I suspect this fix to psnup caused the build problem in > > > <URL: https://bugs.debian.org/1086290 >. Do not know how or how to > > > fix it. > > > > > > The Makefile code in question is generated by Doxygen, so not straight > > > forward to fix in log4c. > > > > The only immediate fix for the FTBFS in log4c I can imagine would be to > > revert the patch for psutils. If you agree I can easily do so ... or > > you can do it as well since finally the package is in debian/ team > > space. > > I've reverted the patch for bug #695178 (which was actually prepared by > upstream - in CC) locally and tried to build log4c against this package. > Since the problem remains I do not think that this specific patch is > causing the problem. > > Reuben, I admit I'm hesitating to bother you with this absolutely > outdated version--you know I'm on your side to get this sorted out--but > do you have any hint for the problem reported for log4c here > > https://bugs.debian.org/1086290 I see that the point at which the build breaks down is: psnup -2 refman.ps >refman.ps psnup: page width and height must be set This happens when the paper size is not set. On Debian, the paper size should default to the system default given by libpaper. It seems that does not happen here. The version of libpaper that I maintain (version 2) is not yet in unstable (only experimental), so no major changes have happened here for many years; however, I see that recent changes have involved the default paper size. I can see that libpaper1 is installed in the full build logs, and crucially there is this line: Creating config file /etc/papersize with new version This is the system-wide default paper size in libpaper, and with this in place, a request for a default size should always return a result. The warning that doxygen gives about the setting of PAPER_TYPE=a4wide seems to be doxygen-specific. So I'm sorry, I don't know what is going wrong; some further debugging will be required. I tried building log4c myself from source, and could not reproduce the error; but the relevant file refman.ps was 0 bytes, which seems like there's another problem. -- https://rrt.sc3d.org
