[calibre]: v4.0 depends on qt5-webengine
calibre v4.0 will be released soon and will depend on 'qt5-webengine' - the current version in parabola will probably be the last release to support 'qt5-webkit' - eli and freemor have done some preliminary work on preparing for the transition; but im not sure what the status is - there was not yet a BR for it; this is a good time to start tracking progress
"the transition" -- nothing really needs to be done for the current version, but since it is a python2 program, it will eventually need to be ported to python3, or else keep python2 in the repos past the time when the PSF officially end-of-lifes the Python 2 release branch, or else drop calibre from the repos.
For various background details on the effort to port calibre to python3, see https://github.com/kovidgoyal/calibre/pull/870
Current status: I can successfully run it, more or less (and the rate of bug finding has dramatically decreased), and the AUR package for the git version actually builds both python2 and python3. But any future fixes to make python3 work will require forking and backporting due to the fact that the upstream master branch is solely based on webengine. Hopefully those won't be too major, though.
Updated by bill-auger 10 months ago
regarding python, calibre is just a drop in the bucket compared to the massive project that obsoleting python2 entails - debian started down that road over a year ago, maybe two years ago, with the aim to purge python2 completely before debian10 - for the largest part, the plan was to simply drop any program that did not port to python3 - in the end, that did not happen, and python2 is in debian, presumably to stay for a few more years - according to archweb, there are nearly 1000 packages that depend on python2 today - does arch have a roadmap or expected deadline for obsoleting python2?
We have no set plans -- we are trying to port all leaf packages to python3, and, for python2 libraries, if nothing needs them then we are dropping them for being "unused packages". There are no firm decisions on what to do in 3 months from now.
Debian's story is very different from Arch or Parabola. Once Buster was released (before the official EOL date) with python2, they needed to support it for the life of Buster. By the time Bullseye is released, January 2020 will have long passed, and they'll have had the chance to drop it (the work is already underway in sid to remove python2 itself).
During the support period for Ubuntu, Red Hat, and other commercial support distributions, it's quite reasonable to expect python2 to get new upstream maintainers, at least for security purposes (and the interpreter is stable enough that very few changes were going to happen anyway). So that's always an option, though really, it depends on how much we value any specific remaining python2 packages. I was committed to keeping calibre in the repos, even if that means preserving python2 until it is ported, but it looks like that won't be necessary anymore -- I'm not sure what the story might be for other leaf packages.
Updated by bill-auger 9 months ago
- Status changed from in progress to forwarded upstream
same segfault on arch32 - both with calibre v3.48.0 and also 4.1.0 - upstream BR: