> Personally I don't see the benefit of 1 big file vs. many 0-length files to justify > the cost (time and complexity) of a PostgreSQL change, with the > corresponding cost of making use of this new functionality based on > PostgreSQL version. > >From a packaging stand-point 1 big file is better than tons of 0-length files. Fewer files to uninstall and to double check when testing.
As to the added complexity agree it's more but in my mind worth it if we could get rid of this mountain of files. But my vote would be the wild-card solution as I think it would serve more than just postgis need. Any project that rarely does anything but add, remove, or modify functions doesn't really need multiple upgrade scripts and I think quite a few extensions fall in that boat. > We'd still have the problem of missing upgrade paths unless we release a new > version of PostGIS even if it's ONLY for the sake of updating that 1 big file (or > adding a new file, in the current situation). > > --strk; I think we always have more releases with newer stable versions than older stable versions. I can't remember a case when we had this ONLY issue. If there is one fix in an older stable, version we usually wait for several more before bothering with a release and then all the stables are released around the same time. So I don't see the ONLY being a real problem.