Theo de Raadt wrote on Sun, Oct 23, 2011 at 12:01:18PM -0600: > Ingo Schwarze wrote:
>> Regarding the brute force approach, good luck backporting the >> build system changes 28 times (from 5.0 to 2.2), demonstrating >> that the backport is correct and not missing any manuals in any >> version, and then finding a machine that is still in working order, >> able to run 2.2, and fast enough to do 28 production builds. >> >> That sure sounds like a horrific task. > Especially when one considers that in the past the tree used to > contain many many more machine-dependent manual pages. Oh right. 5.1 is going to be the first release where all machine-dependent manuals for all architectures will be installed on all other architectures as well, such that there is no longer any risk that you miss whole pages for, say, the binutils of the GCC 2 era just because you happen to read manuals on a GCC 4 architecture. That's one of the nice details we did during s2k11 in Ljubljana (say hi to Mitja!). But wait. In 5.1, the same set of manual pages will be available on each machine. However, the *content* of, say, the gcc(1) manual will still look different on a GCC 2 and on a GCC 4 architecture. Drats, my evil plot for simplicity was foiled again...
