On Sat, Dec 15, 2007 at 12:07:44AM +0000, Simon McVittie wrote: > On Fri, 14 Dec 2007 at 00:53:28 +0100, Michael Koch wrote: > > On Mon, Dec 10, 2007 at 08:40:43PM +0000, Simon McVittie wrote: > > > Michael: how would you recommend proceeding with this? Should I bother > > > asking for a sponsored-NMU of a known-broken package in the hope that it's > > > xmms2's fault, should I add the Recommends, should I file a separate bug > > > for > > > the Recommends? > > > > I would be for a separate bug for the Recommends so the maintainer can > > add it and add some documentation in /usr/share/doc/abraca/README.Debian > > on this. > > Filed (#456369), please follow up there if required. > > > Is using a temporary $HOME really such a good idea? What do you thin > > about just disabling/removing (via patch) the stuff in ./waf that needs > > $HOME? I'm not a python guy, so I cant help much here. > > I've had a look, but I hadn't realised just how bizarre waf is. ./waf is > a self-extracting Python script - it contains some Python code, followed > by a string containing a tar.bz2 file encoded in Base85. It unpacks the > tarball into the build directory as .waf-VERSION/ and loads the > libraries that contain most of its code from there. So, patching it > would be possible, but you'd have to unpack the tarball, patch it and > repack it, or just leave it untarred; either way creates a far larger diff > than is necessary, and goes against how waf appears to be conventionally > used (which is to put a verbatim copy of it in your source tree). > > waf does have a --nocache option which looked promising, but instead of > "don't use a cache in $HOME" it just means "clear the cache in $HOME > before beginning", so that's no help. > > I did spot a silly mistake in my patch - I should have set > WAF_HOME = $(CURDIR)/debian/tmp-waf-home rather than just > debian/tmp-waf-home, although waf doesn't actually seem to have a > problem with its $HOME being a relative path - so please make that change if > you're going to NMU this.
Uploaded finally. Cheers, Michael -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
