2009/2/12 Panu Matilainen wrote: > On Tue, 20 Jan 2009, Rakesh Pandit wrote: [..] > > That's one possible split (but something like tryReadManifest() would be > better name than "check manifest), another possibility might be splitting > out both the package and manifest reading to another function? >
Yeah .. will update. > Which reminds me... rpmgi.c has bits and pieces to do pretty much just that. > AFAIK the idea behind introducing rpmgi.c has been to unify the > file+manifest handling in both install and erase operations, it just was > never finished so we have query code using rpmgi and install code > duplicating a lot of the same. OTOH rpmgi is a dragon of its own :) but > maybe something to look into. Doesn't mean we can't clean up rpmInstall() by > other means first, switching to use rpmgi might well be easier once the > rpmInstall() blob is chopped to smaller pieces. > In case it okay I would like to follow later approach (... fight with two dragons separately ;) i.e chop rpminstall into pieces (even try to simultaneously kill as much as repeated stuff, but preference being to get the complex rpmInstall ease a bit) But once it is done we can kill the duplication easily. -- Regards, Rakesh Pandit _______________________________________________ Rpm-maint mailing list [email protected] http://lists.rpm.org/mailman/listinfo/rpm-maint
