Skip to Content.
Sympa Menu

sm-discuss - Re: [SM-Discuss] SVN as p4 replacement

sm-discuss AT lists.ibiblio.org

Subject: Public SourceMage Discussion List

List archive

Chronological Thread  
  • From: "Sergey A. Lipnevich" <sergey AT sourcemage.org>
  • To: "sm-discuss AT lists.ibiblio.org" <sm-discuss AT lists.ibiblio.org>
  • Subject: Re: [SM-Discuss] SVN as p4 replacement
  • Date: Mon, 20 Mar 2006 16:21:31 -0500

Arwed von Merkatz wrote:
Choice isn't the problem as long as _every_ combination of choices
builds.

I have this in DEPENDS:
depends APR '--with-apr=/usr' &&
depends APU '--with-apr-util=/usr' &&

I would gladly accept a patch to make it easier for the user to make a consistent choice every time this is processed. Extra "point" would be awarded to make this patch only touch the DEPENDS file ;-).

I have never been able to get a subversion install with
apr/apr-util, always had to go for apr0/apr-util0. No idea if this works

All I can say here is that I rarely test the spell with apr0 because I was focusing on Apache 2.1+ lately. I've been upgrading my servers to new versions of Subversion in step with releases (more or less), and never had a compilation problem. I'm biased of course, but my point is that more failure scenarios are needed to understand what exactly is breaking up.

better now, but the subversion spell really breaks a lot - be that due to
apr changes, swig changes or subversion changes.

Again, I am not aware of serious persistent breakage, and neither is Bugzilla. There's three open bugs now (10721, 10427, 6674), and only one of them is "unclear." 6674 is a sub-depends issue; 10427 is the incomplete patch for db-4.4. So the only "difficult" problem is 10721, that I'm going to look into.

A regression test suite could certainly help here, we'd just need a way
to connect those spells. That would maybe also help without a full set

Do you mean a regression test for the spell or for Subversion itself? The former would be great, but how? I haven't seen any reports from Prometheus in a long time, and that's the only regression testing facility I know of. The latter doesn't really help because to be regression tested, Subversion has to be compiled first, and this where the problems start.

So, to summarize, please file bug reports and describe the Subversion breakage you experienced :-)!

Sergey.




Archive powered by MHonArc 2.6.24.

Top of Page