On Tue, 2 Jul 2002, Lars Gullik Bjønnes wrote: > Allan Rae <[EMAIL PROTECTED]> writes: > > | On Sun, 30 Jun 2002, Lars Gullik Bjønnes wrote: [...] > >> So the solution is either to make all generations of this document to > >> use the same sorting rules, _or_ to not have this file automatically > >> genereated, but manually edited/created instead. > > > | _Or_ not have it in CVS. > > That is the worst option since we will have a _very_ hard time to get > a updated lyx.pot that we are confident that are correct. > (currently it is not because of the xforms/forms_* changes)
I'm not sure I understand you here. If we continue to auto-generate POTFILES.in but don't have it in CVS we always keep it up-to-date as we do now (so long as people run make from $top_srcdir -- which they need to to make a dist) *but* we don't have to worry about sort order induced conflicts on updates (like I got yesterday) and so have a simpler handling (no need to make sure every possible 'sort' results in the same order). lyx.pot is regenerated when a dist is made -- such as for a pre-release. Translaters are encouraged to use the dist tarball. If lyx.pot and POTFILES* are included in the dist the translaters use, where is the problem? The only problem we really have now is that sort order varies from OS to OS. How does that affect whether we trust lyx.pot? If lyx.pot is regenerated when a dist is made why would we not trust it? -- especially if the autogeneration of POTFILES.in has been working successfully for some time now. I automated the update of POTFILES.in etc. because there were complaints that doing it from autogen.sh (effectively manually) meant we didn't have a POTFILES.in we could trust -- IIRC one or more of the dists didn't have autogen.sh run first so POTFILES.in and hence lyx.pot (because a couple of files weren't included) were incomplete. Allan. (ARRae)
