On Fri, Nov 15, 2002 at 12:07:45AM +0900, Junichi Uekawa wrote: JU> Yes, I was quite wondering that too, and people tend to disagree on JU> that point, and some people tend to be walking around filing RC JU> bugs on this regard.
RC bug would definitely be an overkill, I have shown that FHS is ambiguous about this issue, so either way (lib or share) doesn't violate anything. JU> Scripts may have some architecture dependency etc., right ? JU> /usr/share is sharable among different architectures as well. Good point: notion that a script can theoretically run on any platform does not guarantee that it will. That is the difference between program and data: you can be able to read, but not to run the script, and program that can't be run is not usable. Aside from pack() and other evil things, it can simply depend on interpreter version that is not available for some architectures, as is the case with Alicq: it depends on tcl8.3 which is not available on Hurd and SH. Not to mention Java "portability": we had problems porting Java applications between Windows and Linux on i386 alone, I can imagine what we'd encouter had we had to support MIPS. -- Dmitry Borodaenko