bug#24404: "calibre" package fails to build.

[Efraim Flashner <efraim@flashner.co.il>]

On Sun, Jul 22, 2018 at 11:49:14AM +0200, Andreas Enge wrote:
> Hello,
> 
> I ended up disabling tests (see comments in the patch).
> Now the package builds, but tries to install into the Qt directory; at the
> end of the cmake phase, it prints:
>   -- Installing in the same prefix as Qt, adopting their path scheme.
> The previous version of the package contains a phase to adapt this,
> but the .pri files to be modified do not exist any more.
> 
> I also tried to inherit from qtsvg like other qt* packages; but the
> gnu-build-system does not work any more for qtwebkit, the build finishes
> in a few seconds creating only the documentation and not compiling the code...
> 
> At this point, I am giving up; it would be nice if someone else could take
> a look, I am attaching the current patch.

Thanks for getting the patch this far! I switched the 'configure phase
from (invoke qmake) to fully cmake with some necessary configure-flags.
It seemed easier than trying to convince qmake to install to the correct
location.

I also left the tests disabled, 7+ hours compiling on my fast aarch64
board was quite long enough. I'm not opposed to re-enabling them but I
don't want to debug failures.

> 
> If there is no progress during the next few days, I would suggest to re-add
> pyqt@5.9 for calibre. What do you think?

After fixing a bug in optipng (bundling outdated copies of libraries is
definately a bug) I was able to compile calibre on aarch64. I run it
headless, so I wasn't able to test it but hopefully it's back to
working.

> Andreas
> 
> PS: There is a thread from 2016 in which the Calibre author states that he
>     will stick with qtwebkit and in the worst case take over the maintenance
>     of a fork:
>        https://www.mobileread.com/forums/showthread.php?t=270258

If he's planning on going the same route as gnucash I assume its only a
matter of time until he realizes distros will drop calibre rather than
carry along his beloved cruft.


-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted