On Wednesday 30 November 2005 20:43, William A. Rowe, Jr. wrote: > I'm voting -1 until the issue of packaging apr/apr-util/apr-iconv in the > httpd tarball is resolved. The last I heard, there were folks voting > AGAINST this, yet I saw these trees in httpd-2.1.10 tarball. Why?
Can someone clarify: what happens *by default* if APR 1.0/1.1 is found on a target machine? If it tries to build against that, I'd support a -1. If it does something sensible - which could be emitting an error message and refusing to build - I'd not worry. > And the suggestion to have an httpd-2.x.x-bundle.tar.gz was raised, that we > include apr/apr-util/kitchen sink. That never saw a resolution, with > several of those against apr being rolled into httpd, also being against > this proposal. No legitimate counterproposals were offered. I diskile bundling APR, and dislike even more bundling third-party libs like expat and pcre. But I thought I/we had just lost that argument with louder voices. > There's no way that this list has agreement/concluded vote on if srclib/ > should include apr/apr-util/expat, and when it's present ./configure is > doing the wrong things. So we perpetuate (nay - it's made worse) the 2.0 > just to push this out the door. Yep. The software is better than 2.0, but the packaging isn't. I don't like it, but since I don't feel willing/able to fix the packaging myself, I just grit my teeth and bear it. > Roy's point of how f'ed up many fink distributions are is rather funny, > it's the reason my Mac isn't building httpd-2.2 from svn, and the reason > I'm building new toolchains on Win32. The last thing I want is for httpd > to be as much of a mess as most of the packages out there, today :-) Nah. If we don't mess up enough, the packagers will do that for us. -- Nick Kew
