I added this patch because the original patches disappeared from the Debian mirror. In fact, it is the same patches below:

https://github.com/jcpetkovich/overlay-petkovich/blob/master/app-text/calibre/files/calibre-2.9.0-no_updates_dialog.patch

https://gitweb.gentoo.org/repo/gentoo.git/plain/app-text/calibre/files/

The reason, as the patch name suggests, is to prevent Calibre to search for updates. I will look at it later.

@AndersonTorres That patch which sets self.update_checker to None doesn't prevent Calibre to search for updates and actually breaks some things:

1. Whether to search for updates or not is decided based on command line arguments. The part of the patch that I left untouched, sets the default value of that command-line argument so that updates are disabled. The command line argument is used as follows (https://github.com/kovidgoyal/calibre/blob/v3.20.0/src/calibre/gui2/update.py#L171):

        if not opts.no_update_check:
            self.update_checker = CheckForUpdates(self)
            self.update_checker.signal.update_found.connect(self.update_found,
                    type=Qt.QueuedConnection)
            self.update_checker.start()

Note that if update checking is disabled by the command line argument, no attribute update_checker is added to self. However, the patch that I removed, did add that attribute and set its value to None.

2. When closing Calibre, these lines are executed (https://github.com/kovidgoyal/calibre/blob/v3.20.0/src/calibre/gui2/ui.py#L975):

        if hasattr(self, 'update_checker'):
            self.update_checker.shutdown()

Note that if self.update_checker is None, this causes a run-time error. The unpatched Calibre version handles this so that update_checker is not an attribute of self when updates are disabled. So there is no need for any patching.

Those are the only two locations that refer to update_checker. Based on this, I'd say the patch can be removed as I did.

Calibre master branch was just modified so that this works even if self.update_checker is None. However, a) it's not yet in a release, b) the patch is still unnecessary.

The reason I asked why this was added is that I don't see any reason for the patch because Calibre works without it, and with it, it actually doesn't work. So I thought there's probably some reason that isn't clear to me. I guess the patch was fixing some issue that had existed in some older version of Calibre.

@jluttine , do you think the patch can be completely erased?

@AndersonTorres I would erase only that part that I've erased in this pull request. The remaining part of that patch disables the update checker by default and offers a command line switch to enable it. That's good. (Normally Calibre does the other way around.)

@jluttine

Now Calibre was updated. Can you solve this conflict and re-submit it? (You can just submit the modified patch you are suggesting.)

@AndersonTorres Yes, I now updated this pull request. It just removes an unnecessary part of the patch. It wouldn't have caused issues anymore on 3.21.0 because of an upstream fix, but it is unnecessary anyway so better to remove so it won't start causing issues at any later point and the nixpkgs codebase stays cleaner.

@AndersonTorres Is this merge-ready?

@srhb Yes! I will merge it now! (Real life issues again...)