control: block 823264 by 824673 Hi,
[ This has already been requested/asked in #823264 (reported against wine). Added to CC now, but please send answers only to #824673. Blocking that bug by this one here, because we usually fix stuff in wine-development first. ] I wonder if it's really necessary to use wine32-tools at all. Or is winegcc from wine64-tools, with gcc-multilib installed, producing the same results if "-m32" is specified? If we indeed need wine32-tools: For the dependeny on gcc the following had been suggested: gcc | gcc-multilib:amd64 [i386] I successfully tested this here, but I'm not sure if specifying a specific package arch this way (":amd64") is allowed. Further gcc-multilib has no Multi-Arch field, while I'd expect "m-a:foreign" or at least "allowed" for this to work. I'll ask the multi-arch folks about that soon. Is it really necessary to explicitly specify "-m32" for compiling 32-bit apps on amd64 with wine32(-development)-tools? If yes, I wonder if this should be mentioned either in the README, or added with some logic to the winegcc wrapper script. We also discussed a possible multi-arch dependency on perl in #823264. I just noticed in dh_perl(1) that ${perl:Depends} gets resolved to perl *or* perlapi, while no other alternative value is mentioned there. So we may use: ${perl:Depends} | perl:any [i386] We may even use "perl:any" only, but I'd like to keep the variable in order to notice if anything changes there. Some way for dh_perl to handle this would be preferable. Or we depend on perl:any and recommend ${perl:Depends}. However I'd like to first know for what exactly perl is needed, and if perl:any would work for wine, or if native (i386) perl is needed for at least some things. Anyone? Greets jre