El Thu, Jun 27, 2019 at 04:07:53PM +0200, Agustin Martin deia: > On Fri, Jun 21, 2019 at 07:24:15AM +0200, Xavi Drudis Ferran wrote: > > > > Hello. Maybe you know already (because the problem seems to be lack of > > maintainer). But there is a python-uniconvertor 2.0 that no longer depends > > on python-imaging but on python-pil (python2, I believe) > > > > https://sk1project.net/modules.php?name=Products&product=uniconvertor&op=download > > Hi, > > Even if that dependency is declared, python-uniconvertor in Debian does > not really depend on python-imaging, but on a copy of it embedded as > sk1libs.imaging in sk1libs, which is not in Debian. > > The biggest problem is the lack of sk1libs, because there are some other > functions that are required by uniconvertor as soon as you try to go > beyond the usage message (See https://bugs.debian.org/820748). > > I am not a python expert, but for the imaging part I would expect something > similar to attached patch to work. It changes dependency to python-pil and > tries to change calls to sk1libs.imaging stuff to the equivalent PIL calls. > > Regards, >
Thank you. I'm looking into 2.0rc4. It seems to use PIL already, but it uses a previous version of reportlab (should work in stretch but not buster, I think it needed python-reportlab < 3.4.2 IIRC). I only tried generating an pdf from an svg and failed. So I suspect 1.1.5 would not work with buster python-reportlab either. In 2.0rc4 there are also generated files in src/uc2/cms/*_rc.py that I don't know how DFSG compatible are. I found the script that generates them from icc files, and some carry public domain or other licenses, but I don't know what are the original icc profile files used to generate the _rc.py files. Meybe one could use icc files from icc-profiles-free or other packages. But how do I know if I'm breaking something with the change? I mean I could maybe with some effort put up some version of 2.0 that works for me (or not, it's not really my stuff tbh) but I would not know how to test it all. upstream offers 2.0rc4 debs for 2 architectures, but for debian 9, not 10. That would confirm my dependence issue. But I'll keep your patch in mind. Maybe it's easier to port 1.1.5 to buster (or sid). Or maybe we should just ask upstream for their plans.
