Hi Nikita, > > Hi Nikita, > > > > > Probably I will need one more dpkg-cross upload - will you sponsor > > > me? > > > > Of course! > > As a e-mail debug session with Erich Waelde <ew.ng116837@online.de> > shows, there is some problem with umask handling - looks that in some > situation, dpkg-cross's umask setting does not help and package is > built with files that are not world-readable. I need to look deeper > into this to find out what is going on. > > Also I want to (partially-)close some bugs - e.g. #284439 - I have > some ideas about how more or less general framework for > autotools-vars discovery should work, but I could not find time to > implement this for some months, so it is better to include proposed > config.m32r at least. > > I *hope* to prepare an updated package this week. > > > > > > Until now, I used to use update-alternatives instead. Is that bad > > > for some reason? > > > > Ohh, I did not realised that you used update-alternatives. Sorry, > > but I worked on gcc-2.95 cross-compiler for PowerPC. Now it works > > with the brunch version of dpkg-cross's APT wrapper. > > > > The usage of update-alternatives within the gcc packges is a 80% > > Why not 100%? > > I think that alternatives is better here, because it should be > developer's chose which cross compiler version is default for him, > not mine or Debian's - because only developer knows which works > better for his target. > > However, maybe I don't see something? Please explain your reasoning. For C++ you *have to* use the Debian's default compiler due to ABI incompatibilities for the different GCC versions. The ABI was changed multiple times for different architectures. If a developer uses his favorit compiler it may not work for C++ in Debian because the precompiled C++ libraries used an another compiler version. -- Raphael Bossek
Attachment:
pgp_ldVtnz0fW.pgp
Description: PGP signature