Hi, > > The only reason that pyxplot build-depends on libmagickcore2-extra is to > > get svg to png conversion functionality in imagemagick. Since pyxplot > > doesn't actually use any symbols itself from imagemagick at all (it just > > calls convert!), it shouldn't need to worry about transitions in helper > > utilities that it calls. [..] > Stupid question: wouldn't graphicsmagick (and > graphicsmagick-imagemagick-compat) be sufficient for your use case?
A good thought as that would at least side-step this dependency problem, but no, graphicsmagick does not correctly convert the svg to png in this case so it's not a suitable solution. So back to plan A... since I'm not linking against symbols in libmagickcore{2,3}-extra at all, it's probably wrong for me to have to depend on the package at all. That the soname change has affected pyxplot in this way illustrates the point... getting the right library for imagemagick to be able to manipulate svg files should be the job of magick packages and not other unrelated packages. If I'm not linking against the library then I shouldn't have to worry about its soname. The wrong person is having to worry about these details; the abstraction isn't working properly. The suggestion was made on IRC that ideally, there should be a binary package imagemagick+extra (or something), which would depend on the libmagickcoreN and libmagickcoreN-extra packages for the correct N to provide the functionality required. Thoughts? cheers Stuart -- Stuart Prescott www.nanoNANOnano.net
signature.asc
Description: This is a digitally signed message part.