On Wednesday, April 01, 2015 10:31:59 PM Andreas Cadhalpun wrote: > Hi Sebastian, > > On 01.04.2015 22:16, Sebastian Andrzej Siewior wrote: > > On 2015-03-30 23:09:32 [+0200], Andreas Cadhalpun wrote: > >> On 30.03.2015 22:17, Sebastian Andrzej Siewior wrote: > >>> On 2015-03-27 17:45:04 [+0100], Andreas Cadhalpun wrote: > >>>> Yes, it's probably best to hide these implementation details from the > >>>> API/ABI. But then the function declarations should be moved from the > >>>> installed tfm.h header to a private one, e.g. tfm_private.h. > >>> > >>> Hmm. Right. I will check with upstream what he thinks would be best > >>> practise here. > >> > >> OK. > > > > so upstream asked if we could do the symbol hiding within our debian/ > > folder. Doesn't look that bad, does it? > > I don't think it's a good idea to perpetually deviate from upstream's > ABI/API, because then projects might start using some of this low level > stuff and not work in Debian. > > Does upstream think this low level stuff could be useful outside of > tomsfastmath or where does the reluctance to hide these symbols come from?
I think it would be good to get all this sorted out before we upload to the Debian archive. I definitely don't want to have a Debian specific ABI break after it's in the archive. Scott K _______________________________________________ Pkg-clamav-devel mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-clamav-devel
