>There are far bigger fish to fry; I don't imagine this will be a priority >for me before significant headway is made in the archive wrt spurious >dependencies on libs that *don't* have stable ABIs.
If you do this, it will make it easier for me to *find* the most problematic of those libs so I can write patches for them. :-) I have been randomly working through lib* packages listed in checklib. The checklib output (http://rerun.lefant.net/checklib/) is polluted by a large number of useless dependencies on zlib (and a few other "nonproblematic" libraries), which can make it rather hard to sort through and find the "really problematic" useless dependencies. Many of the useless dependencies on zlib come from freetype, because freetype is *exporting* the bad zlib dependency to all its users. If some of these *heavily exported* useless dependencies were eliminated, it would make it a lot easier to see the actual scope of the problem with relation to libraries with unstable ABIs. A single useless dependency on a package with a stable ABI is no big deal; dozens or hundreds makes it hard to find the real problem cases. (I'm not currently prepared to write a program to reparse the data from checklib, but if bugs like this fester, I may have to waste a few days doing so.) It's like being lintian-clean. It's certainly of "minor" importance, but leaving it lying around is untidy and annoying; just do it next time you make an upload (after etch releases and testing unfreezes, that is, of course). If you had a lintian error in your package, would you make ten uploads over the course of a year without bothering to fix it? Of course you wouldn't. But that's what's been done with this bug. I think "need to deal with it" is absolutely correct. "Need to deal with it urgently" would be wrong. -- Nathanael Nerode <[EMAIL PROTECTED]> Theocracy, fascism, or absolute monarchy -- I don't care which it is, I don't like it.
pgpSOH1AbFxhz.pgp
Description: PGP signature
