On Sat, 20 Nov 2010 15:16:15 +0100, Petr Baudis <pasky@suse.cz> wrote:
I'm aware of this bug. If the memcpy() backwards copy performance gain was significant,
Where do you draw the line? Some programmers also relied on undocumented behaviour or internal variables and cried murder when that changed. I remember times where some programmers assumed they could write to string constants.Or take the case of rpm segfaulting because it was linked statically but used one of the name resolution functions, i.e. those supplied by the libnss* libraries. Had that segfaulting binary been flashplayer or acrobat would you have changed glibc (which would only been possible by completely ripping out the nss stuff) or told the issuer to fix his application? Do you really believe the situation would have been different had the glibc folks warned the software world in advance that this would be coming? I've seen enough stupid programming bugs in open source applications to believe that binary only software isn't better but even worse. The question remains, where do we draw the line of acceptable wrong coding? Philipp -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org