Hi, For our project, we are using msgmerge --update with the --sort-by-file option to update the language po files from the pot file generated by xgettext. Now the po files have comments like:
#~ msgid "Password" #~ msgstr "Parool" which contain msgids not used anymore. The problem we now have is that every time we run msgmerge, it will reorder all these commented out msgid-msgstr pairs. This generates a hugely bloated diff. Because of this, we run xgettext+msgmerge very rarely over our project, because we'd like to keep the git repo size small. But that doesn't help our translators, because they would like to translate the texts very shortly after they got introduced in the codebase. Therefore, we would like to run xgettext+msgmerge more often, but with hopefully avoiding to bloat the git repo. I can partially work around the problem by filtering the patch for lines starting with #~, but that would mean the list of commented out translations can't grow, which is sad, and it also eliminates the actual purpose of the #~ lines. I haven't tested whether the --sort-output param solved this problem, but I would prefer to not use it, in order to keep msgids with a similar context together. Can msgmerge somehow be made "stable", in the sense that it doesn't reorder these commented out translation pairs? I'm using msgmerge version 0.19.2. Thank you. est31
