> Hi Gürkan, Hi Tobias, >> colmap (3.6+really3.6-1) unstable; urgency=medium >> . >> * New upstream version. (Closes: #957094) > > (I wonder, if is fixes by the the upstream bug; it is still marked "open"… > but maybe upstream did not track this properly) Indeed, I should close the bug, once it really is closed in Debian. https://github.com/colmap/colmap/issues/858 (Alex is my aliasname, to get rid of non-ascii characters in my fullname) > However, unfortunatly it fails here to build in a sid pbuilder environment, > I'm attaching the full log, seems so that there is something with ceres. > (did not investiage further) I confirm. > There are two undocumented changes: > - Rules-Requires-Root > - the commenting out of the overriding for dh_autotest. > > Especially for the last one, the "why this change is here" > would be quite interesting; possibly this should close #909198, > woudn't it? But without the reason in d/changelog, its just guessing. True, I have fixed all issues except the minor dep3 headers one in the source package at: http://phd-sid.ethz.ch/debian/colmap/2020c/colmap_3.6%2Breally3.6-1.dsc > Some other minor stuff: > - the patches have no correct dep3 headers. > - quite old compat level, not using debhelper-compat > - you could update copyright years, at least for the debian/* part > > Please take a look and let me know when it's ready. Will update when it's ready. PS: background info on the building problem from #debian-devel 07:56 < tarzeau> could i be hit by #883619 again? according to #971889 ? 08:39 < _jwilk> tarzeau: CeresConfig.cmake has set(CERES_EIGEN_VERSION 3.3.7), but there's only libeigen3-dev (>= 3.3.4) in Depends. 08:41 < _jwilk> Looks like it's only happy with this exact version: "Found Eigen dependency, but the version of Eigen found (3.3.8) does not exactly match the version of Eigen Ceres was compiled with (3.3.7)." :-/ 08:43 < tarzeau> _jwilk: yes :-/ what am i supposed to do? 08:44 < _jwilk> Curl up into a ball and cry? 08:44 < tarzeau> (currently rebuilding eigen then ceres to confirm, takes hours) 08:44 < jochensp> also #860309 08:44 -zwiebelbot:#debian-devel- Debian#860309: libceres1: Requires rebuild for Eigen 3.3.2 - https://bugs.debian.org/860309 08:45 < jochensp> ..and #868355 08:45 -zwiebelbot:#debian-devel- Debian#868355: nmu: ceres-solver_1.12.0+dfsg0-1+b3 - https://bugs.debian.org/868355 08:46 < jochensp> that one discusses if the check is too strict, maybe we should follow up on that 08:47 < _jwilk> Regardless of whether it's too strict or not, Depends must match what is being checked. 08:51 < tarzeau> jochensp: thanks! waiting (fixing it formyself meanwhile) 08:52 < jochensp> agreed but how would it look like? libceres-dev depending on $Version of libeigen3-dev and triggering a transition for every new version of eigen3? 08:55 < jochensp> _jwilk: ^ 08:57 < _jwilk> Yeah. Better to trigger transition with unsatisfiable Depends than with FTBFS, like it happens currently. 08:58 < jochensp> ok, I can probably add some version magic later today 08:44 -zwiebelbot:#debian-devel- Debian#860309: libceres1: Requires rebuild for Eigen 3.3.2 - https://bugs.debian.org/860309 08:45 < jochensp> ..and #868355 08:45 -zwiebelbot:#debian-devel- Debian#868355: nmu: ceres-solver_1.12.0+dfsg0-1+b3 - https://bugs.debian.org/868355 08:46 < jochensp> that one discusses if the check is too strict, maybe we should follow up on that 08:47 < _jwilk> Regardless of whether it's too strict or not, Depends must match what is being checked. 08:51 < tarzeau> jochensp: thanks! waiting (fixing it formyself meanwhile) 08:52 < jochensp> agreed but how would it look like? libceres-dev depending on $Version of libeigen3-dev and triggering a transition for every new version of eigen3? 08:55 < jochensp> _jwilk: ^ 08:57 < _jwilk> Yeah. Better to trigger transition with unsatisfiable Depends than with FTBFS, like it happens currently. 08:58 < jochensp> ok, I can probably add some version magic later today (with their written permission) > Cheers, >
Attachment:
signature.asc
Description: OpenPGP digital signature