Hi, On Tue, Feb 18, 2025 at 1:55 AM Maxime Ripard <mripard@xxxxxxxxxx> wrote: > > > I also questioned whether this really made sense to try to do with a > > Coccinelle script and I still don't think so. It looks like Dmitry has > > already reviewed the first few of your patches and has repeated my > > advice. If you want to help with the effort of addressing this TODO > > item then that's great, but I'll stop reviewing (and start silently > > deleting) any future submissions of yours that say that they're done > > entirely with a Coccinelle script unless you address this point and > > convince me that your Coccinelle script is really smart enough to > > handle all the corner cases. I'll also assert that you should review > > Tejas's submissions to see how these conversions are expected to go. > > I couldn't find that in your first answer though. What corner cases do > you have in mind, and why do you think coccinelle can't handle them? My gut says that it is a difficult problem to make this kind of change purely with Coccinelle. That's not to say I couldn't be convinced if someone gave some good evidence showing some amazing patches generated by a Cocinelle script. To show this, I would expect someone to understand what Tejas has been doing and then compare that to what the script can produce. In theory, you could even run the script on an old version of panels (before Tejas's fixes) and compare what the script does. If you can make the results nearly the same then that's amazing. > Also, why do you think ignoring a contributor after a second mistake is > a reasonable reaction? Just to be clear, I said that "unless you address this point and convince me...". My problem was that I brought up the questions of the suitability of Cocinelle for this problem in response to v1. Then I saw v2 posted without any reply to my concerns and with the same types of problems. I was simply saying that if a v3 was posted in the same vein then I would ignore it. I'm more than happy to have a conversation, but if I start getting new versions that ignore previous feedback and no response saying why feedback was ignored then I will start ignoring new versions. That feels pretty reasonable to me.
