On 08/19/2011 03:34 PM, Jakub Wilk wrote: > * IOhannes zmölnig <zmoelnig@iem.at>, 2011-08-19, 11:30: >>> pyassimp.errors.AssimpError: assimp library not found >> >> hmm, it worked for me...:-( >> trying to find words to ask you how and what, i think i found the >> reason: pyassimp looks for /usr/lib/libassimp.so which is only >> provided by libassimp-dev (which i most likely had installed when >> testing) >> since the findmechanism in pyassimp for dlopen() is rather simplistic, >> i guess this is the reason why it fails. >> >> thanks for pointing that out. > > Yeah, the whole search_library() function could be replaced by a simple: > > return ctypes.cdll.LoadLibrary('libassimp.so.2') > > (Well, modulo support for non-UNIX-y operating systems.) yes, i came up with something equivalent (though not checked in yet), but also using ctypes.util.find_library('assimp') to resolve "libassimp.so.2" dynamically. since i am no python programmer, i wonder whether this is portable (w32, osx) > >>> Also, why did it print stuff on stdout? :/ >> >> ask upstream :-) > > ... no, seriously. > >> Q: as a packager, am i supposed to fix such things? > > At least your are supposed to forward bug reports upstream. (Of course, > your sponsor might choose not to upload the package until a bug that is > particularly severe/annoying is fixed.) sure, but how does this apply here? the python module prints something to stdout when used. is this a bug? what are the guidelines to decide whether i should "fix" a given behaviour that will not do any harm? (right, i can easily think of an example where the behaviour could be problematic. but then i could also think of such things for many other applications and that does not neccessarily make them all buggy) fgmasdr IOhannes
Attachment:
signature.asc
Description: OpenPGP digital signature