From kde-mac Tue Jun 22 05:43:53 2021 From: Ian Wadham Date: Tue, 22 Jun 2021 05:43:53 +0000 To: kde-mac Subject: Re: Install presence Message-Id: <9EAB62DD-3075-4C3E-B34E-63F86908C090 () gmail ! com> X-MARC-Message: https://marc.info/?l=kde-mac&m=162434064328506 Hi Ren=C3=A9 and Aleix, > On 22 Jun 2021, at 6:57 am, Ren=C3=A9 J.V. Bertin = wrote: > On Sunday June 20 2021 20:53:55 Jeremy Whiting wrote: >> All of the above is based on the idea that yes, we want KDE = applications to >> work well on MacOS. I'm not sure if that's the consensus or not. >=20 > I just remembered an episode a few years back. I was discussing a Mac = tweak of on of the plasma components on the plasma-devel ML (could have = been DrKonqi, could have been the platform plugin, or ) and one of the = key plasma devs said something tell-tale like "I think I'm going to = break that, just because I can". I presume it's still easy to elicit a = knee-jerk reaction there by invoking the idea of supporting a relevant = set of KCMs on other platforms=E2=80=A6 I think this may have been to do with DrKonqi, which was not working on = Mac OSX and probably never had worked there. In the second and third quarters of 2014, I was working on fixing = DrKonqi and KCrash. The end-result of the problems was that users of KDE = apps on Apple Mac could not report crashes to KDE developers. DrKonqi = either never appeared on Apple Mac or died immediately or failed to = reach the dialog for submitting a bug on bugs.kde.org. Hence very few = bug reports were reaching KDE developers from Apple Mac, even though = there were lots of bugs in the KDE code for Apple Mac. You may have visited the Plasma devs about this, Ren=C3=A9. During the = transition from KDE 4 to KF5, they unilaterally decided to move the = source code of DrKonqi from KDE 4=E2=80=99s kde-runtime to somewhere in = the KF5 Plasma source-code tree, thus making it more-or-less = inaccessible for porting to other platforms. I hope it has been returned = to somewhere more suitable now. @Aleix: I believe you and Aaron Seigo may have corresponded with Ren=C3=A9= and me about this. Your names also appear in the reviews on Reviewboard = of my bug-fixes for DrKonqi. Cut to the chase=E2=80=A6 = https://mail.kde.org/pipermail/kde-core-devel/2014-October/084208.html, = "Review Request 120431: Fix and future-proof Dr Konqi security methods = on Bugzilla=E2=80=9D, is a thread extracted from Reviewboard. In July = 2014, Bugzilla changed to its version 4.4.5 on bugs.kde.org. At one = stroke, this crippled DrKonqi on all platforms and in all versions of = KDE source code: KDE 4 and KF5, on Linux, Windows and Apple OSX, because = the method of validating DrKonqi=E2=80=99s access to bugs.kde.org had = changed. See also bug report https://bugs.kde.org/show_bug.cgi?id=3D337742= from July 2014. With a little help from Ben Cooksley, I had been able to develop a patch = for DrKonqi that would retrieve and test the version number of Bugzilla = and then use the connection protocol required by that version. The patch = was also written so as to be forward-portable to KF5=E2=80=99s DrKonqi, = with one or two changes of line numbers. I had tested the patch thoroughly on Apple OSX, using a test database = for bugs.kde.org, provided by Ben Cooksley. Now read the review. I get nit-picked to death! No way I wanted to try = and work with those guys any more after that. In the review, it is only when the realisation dawns on KDE people that = KF5 will-be/is-already failing that I get some serious reaction. But = still not much action. In the end, I had to threaten to do my own = =E2=80=9CShip It=E2=80=9D, to be sure to get the fix into the next = bug-fix release of KDE 4, which was just a day or two away. But then = Albert Astals Cid released the change for me, with just a few hours to = spare, and someone else forward-ported the same change into KF5, as I = had designed it to happen. I also had had two earlier reviews, of patches to correct the many bugs = in the Apple OSX implementation of DrKonqi and Crash, and had similar = experiences. At least one of those bugs also applied to Linux, Windows, = KDE 4 and KF5. Then the above-mentioned bug report dragged on till after Christmas = 2014, because of another bug that appeared only in LINUX versions of = DrKonqi after Bugzilla changed versions in July 2014. It was fun trying = to debug THAT on Apple OSX, I DON=E2=80=99T think! I had to get = end-users to generate logs for me. Again there was almost no help from = KDE core developers, but at least they did forward-port my fix for that = problem into KF5 again. @Aleix: Good luck with your talk and everything=E2=80=A6 :-) Cheers, Ian W.