| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| skip_mphe | mphe uses sse which is not available on all archs Disable in makefile the use of this lib as not mandatory | Olivier Sallou | no | |||
| use_debian_libs | use Debian libs instead of embedded one | Olivier Sallou <osallou@debian.org> | no | |||
| run_pipeline | use tools in path instead of local dirs as upstream | Olivier Sallou <osallou@debian.org> | no | |||
| add_hardening | add hardening | Olivier Sallou <osallou@debian.org> | no | |||
| gcc-5.patch | Fix type of prefix_trashable since gcc-5 is picky about it | Gianfranco Costamagna <costamagnagianfranco@yahoo.it> | no | debian | 2015-09-01 | |
| libgomp.patch | Add missing -lgomp | Andreas Tille <tille@debian.org> | no | 2016-12-21 | ||
| spelling.patch | Fix spelling | Andreas Tille <tille@debian.org> | no | 2018-10-10 | ||
| no_std_hash_redefinition.patch | remove redefinition of ‘struct std::hash<__int128 unsigned>’ This definition is not needed anymore with recent compiler versions apparently on _LP64 systems. I'm not sure yet of The proper approach to address it in a more architecture independent way. |
Étienne Mollier <etienne.mollier@mailoo.org> | no | debian | 2020-10-13 | |
| gcc10.patch | fix common gcc10 FTBFS issue After fixing #957519 with no_std_hash_redefinition.patch, a few more issues common to Gcc 10 cropped up. This patch addresses them. |
Étienne Mollier <etienne.mollier@mailoo.org> | no | debian | 2020-10-13 | |
| check_memory.patch | check memory of OAHash construction This has been written to try debugging why autopkgtest failures occurred on 32bit architectures. While the package ended up being removed out of 32 bits architectures repositories, this sort of control may remain useful to users of mapsembler2 to diagnose issues; this is more useful message than "Segmentation fault". |
Étienne Mollier | no | 2020-10-15 | ||
| cross.patch | mapsembler2 FTCBFS: multiple reasons mapsembler2 fails to cross build from source, because it builds for the build architecture. It does not pass any cross flags to cmake nor any cross tools to make. In general, these kind of issues can often be solved using dh_auto_*. In this case, that mostly does the trick except for one aspect. Some makefiles use CC to store a C++ compiler. During cross compilation, dh_auto_build passes a C compiler via CC and that doesn't go well. I recommend changing the variable name to CXX, which is what everyone uses. |
Helmut Grohne <helmut@subdivi.de> | no | debian | 2020-11-09 | |
| gcc11.patch | fix ftbfs with gcc 11 Starting with Gcc 11, ordering relations against pointers are now flagged. This is especially useful to spot iterators with types mismatches compared to the CPU architecture they run on. In this particular case, an iterator is used on a couple of occasions, and tested against with "less than 0" comparisons. In a few situations the absence of typing might have caused issues already, because some occurrences are cast to "long int". In order to match the CPU architecture size, and avoid architecture specific bugs, the proper type would more probably be ssize_t. This type is cast where needed. |
Étienne Mollier <emollier@debian.org> | no | debian | 2021-07-12 |