On 5/3/2018 2:45 PM, Ivan Pozdeev via Python-Dev wrote:
Let me express things right from the horse's mouth.
Ditto, as the only person who responded on the tracker before you posted here and the only person other that Guido to respond on the tracker since and the only person to collect data which mostly refute your claims.
The sole purpose of the tone was to not let the mesage be flat-out ignored. I had my neutral-toned, to-the-point messages to mailing lists flat-out ignored one too many times for reasons that I can only guess about.
For 24 years, from 1994 to 2017, _tkinter.c has gotten 333 revisions at a steady rate, an average of about 1 patch per month. Tkinter's .py files have gotten even more -- 426 -- including 5 in the first 4 months of this year. I think at least 20 different core developers, including me, have been involved over the years. Calling this 'indifference' is dog poo. I believe these patches mostly or all happened from people making relatively calm reports, like you originally did. Other people wait at least a month, or two or three, without *any* response to a patch. If they care, they write a polite note to core-mentorship or pydev. Then someone nearly always responds and reviews their patch.
This time, the situation was too important to let that happen.
The situation was that the tkinter maintainer, Serhiy Storchaka, and a tkinter user, me, had already given you initial responses. I indecated that I considered the issue real and Serhiy promised a review of your patch. The tkinter situation is that if one makes tk calls to the main thread from other threads when using non-thread tcl/tk, bad thinks can happen. No surprise. It turns out that some of the people above have tried but only partly succeeded in making thread calls work even with non-thread tcl/tk. This was generally known, but you added the fact that tcl has a thread compile switch. I do appreciate knowing a concrete reason why different people get different results with the same code. If your patch improves the situation great, but very few people will be affected. This is really not important to people who don't do the above, and not worth disrupting pydev. You appear to claim that your 2nd example, sending fake events to the main thread, reposted to https://bugs.python.org/issue33412, fails even with thread tcl/tk. However, when I tested with 3.6 repository, 3.7 installed, and 3.8 repository builds, it nearly finished rather than failing immediately. When I fixed the deadlock bug, IT RAN TO COMPLETION. TO EVERYONE ELSE: If you want to be helpful, ignore the pydev threads. If you have a tcl thread build, as indicated by the 't' in 'tcl86t.dll' in /dlls, test the example, with my fix, on your machine and report on the tracker.
Whatever anyone may think of this, it worked.
This makes me want to scream. I was about to respond to msg316087 on https://bugs.python.org/issue33257 when I saw the crap about removing tkinter. I considered washing my hands of the issue then and there, to avoid 'rewarding' your nastiness. But because I care about tkinter, I decided to continue on the tracker and then inform you in yesterday's response on the thread that I did so in spite of, not because of, 'this'. Since you ignored me and continue to defend 'this', I say it again, and request that you retract your 'delete tkinter' post. -- Terry Jan Reedy