On Monday 19 March 2007 5:11:49 pm Andre Poenitz wrote: > In some very successful attempt to waste some time I run 'make rpmdist' > and now have a few questions (apart from the obvoius 'Is it really > supposed to fail?' and possibly 'Why?') > > 1. Why contain 3781 out of 7528 file names in the .tar.gz the pattern > '/.svn/'? Is it really necessary to pack the whole svn text-base > on top of the 'real' files?
Probably a leftover from the previous cvs experience, it is a bug. > 2. Why is that process 'pack sources, unpack, compile, break, retry' > used instead of 'compile until succeeded, pack interesting > stuff only'? Please indicate your answer below: > > [ ] Because it was easy to implement on top of the 'dist:' target. > > [ ] Because it is the way we like it (i.e. extensive usability > studies showed that this is the way that hurt most without > any visible gain). > > [ ] For some undisclosed technical reason that I will never be told > about. To simulate a clean build without any contamination from the surrounding environment. Also it should be dependent on a pristine package source, that should be package coming from make dist. After all that is the whole idea of an rpm project. > Of course I am asking to gain some secret knowledge for my "Project At > Work" (tm). LyX takes around four minutes to inform me that it can't > find qmake when running 'make rpmdist' (which, of course, is found by an > ordinary 'make'). OTOH, my PaW(tm) happens to build .rpms in half a > minute or so, so I am obviously doing something wrong there. Building > .rpms is suppopsed to hurt, isn't it? Sometimes. ;-) I am the maintainer from several packages in Fedora so I can speak from my self experience. :-) > Andre' -- José Abílio
