[15:26] OvenWerks: Phasing on studio-controls in jammy has stopped due to multiple crash reports like this: https://nextcloud.ericheickmeyer.com/index.php/s/RJZAjwHdAwef8iR [15:26] (that's a PDF) [22:28] Eickmeyer: so the latest change from python3.8 to 3.10 is broken? [22:29] maybe not but that would be my first take on things [22:31] Eickmeyer: makedirs makes ~/.config/autojack/ and that is non-existant? [22:38] Or python3.10 is pickier? To be honest I am not even sure the os.py module is one of the modules changed at this point or if it is the main python3.10. [22:39] Hmm this is proposed? so the one with the patch in it? [22:41] I don't recall that being anywhere near the patch, do we have these same reports from the release version? [22:57] OvenWerks: Python 3.10 must be pickier. BTW, I'm seeing that error regardless of the version, so it's not actually a regression because of the patch, just got alerted to it because of the SRU. [22:57] can you replicate it on your machine? [22:59] you should be aware that 3.10 is a LOT pickier with stuff now [22:59] it's got a lot of things that cause failtobuild or such stuff that worked fine on 3.9 [22:59] I would try adding global new_name and global config_path 7 lines up [22:59] enter any number of current bugs against packages that ultimately point at 3.10 failures [23:00] * enter any number of current bugs against packages that ultimately point at py 3.10 introduced problems [23:01] ThomasWard[m]: I take it the standard take any variable as global unless set locally has gone away? [23:01] E:Unknown [23:01] i haven't dug into the context in this case [23:01] or tested thoroughly beyond my few global variable cases I have in a 3.10 script that work also in 3.9 [23:02] and in my cases, the variables I was working with were module level variables defined in an `if name == "main":" block and stuff was not behaving with it otherwise [23:02] I can't reproduce here so I am just thinking that is the only thing that makes sense [23:02] * and in my cases, the variables I was working with were module level variables defined in an ```if name == "main":"``` block and stuff was not behaving with it otherwise [23:03] * and in my cases, the variables I was working with were module level variables defined in an "if __name__ == "__main__"" block and stuff was not behaving with it otherwise [23:03] * and in my cases, the variables I was working with were module level variables defined in an if __name__ == "__main__" block and stuff was not behaving with it otherwise [23:03] What it comes down to is that we're seeing a lot of error reports here: https://errors.ubuntu.com/?release=Ubuntu%2022.04&package=studio-controls&period=week&version=2.3.1-0ubuntu2 [23:03] which is what stopped phasing. [23:04] i assume you have specific reports which you think are python 3.10 caused [23:04] (you might not have access to the error tracker) [23:04] because PermissionIssue issues go beyond Python [23:05] @Eickmeyer i see public error reports here, I"m in error reporters i believe due to high level elevated access [23:05] I'm not sure what's causing any of it. [23:05] well for the ones I do see [23:05] @teward001 That was directed at OvenWerks. [23:06] ah check [23:06] @Eickmeyer well from what I can tell here [23:06] well if we make a path based on a path that includes ~/ and then add a path extension creating a path to something missing that original path may get a permission problem as it may be trying to create a directory in system teritory [23:07] that's an accurate assessment [23:07] @Eickmeyer is there a *specific* error report we're looking at right now that's Python driven? [23:07] Typically, in shell scripting at least, it's safer to use ${HOME} as opposed to ~, but I'm not sure that's relevant here. [23:07] because some of the tracebacks here look like it's a problem with how autojack is *written* [23:08] and that they need to update how they write their stuff to be compliant with newer PEP8 and coding requirements [23:08] Right, which is why I'm always first to blame Python 3.10. [23:08] And its newer requirements. [23:08] so if global setting has to now be in every method. [23:08] And keeping up with said requirements. [23:08] > PermissionError: [Errno 13] Permission denied: '/nonexistent' < this one's self-explanatory [23:08] something's trying to create /nonexistent and can't [23:09] and that error actually occurs twice (duplicate error reports, same general evil) [23:09] JSON decode means it was expecting a value and got none or empty [23:09] as the path /nonexistent is not in the code it must be python3.10 creating that result from something else [23:10] yep. i'd offer to assist but I don't know jack about jack (#ILLEGALPUNDETECTED) [23:10] *shot* [23:10] ErichEickmeyer[m: which is the problem with a NUMBER of universe packages. [23:10] s/./ right now/ [23:10] torbrowser-launcher, some Lubuntu stuff, etc. all because Py 3.10 [23:11] Yeah, I've had to deal with a lot of breakage RE: FTBFS due to Python 3.10. Carla, for instance. [23:11] it does however look like some of these error reports are "Not New" and had 20.04 original origin points [23:12] yeah more involved things explode under Py 3.10 when they aren't up to code spec [23:12] i'm digging into torbrowser-launcher's code now to fix stuff for that one but that's probably an easy one [23:12] unlike jack and stuff [23:13] Well, autojack just runs some dbus and jack execution stuff, so it's not that hard if you know python. At this point, I'm sure it's syntax cleanup. [23:13] It's not like it's running a whole DSP. [23:13] good point [23:14] *inb4 things totally explode even worse* [23:14] HEHEHE [23:17] well you KNOW it could happen so I mean :P [23:17] and usually does during the early stages of a dev cycle [23:20] Right. [23:20] Well, feel free to browse the code. lp:studio-controls, make patches, do what you have to do @teward001 . I mean, little errors, but I wouldn't touch jack commands or jackdbus stuff. [23:21] The reason I'm directing you there as opposed to github is because that's got what's in Jammy and Kinetic. [23:25] github has moved on a bit... [23:25] Quite a bit, I imagine. [23:26] convert-studio-controls no longer exists for example [23:29] Wow, good game! [23:33] *accidentally erases huge portions of code* oooooops :P [23:33] @Eickmeyer i assume git :P [23:33] @teward001 Yes. [23:34] just making sure ;) [23:34] (though much of the code does) [23:34] Bzr is just.... bizarre. [23:35] * OvenWerks has forgotten all that stuff [23:35] ThomasWard[m]: And that's why git revert exists. :P [23:35] *reverts to revision 1, force-pushes, breaks revert* ops :P [23:36] nah i'm not destructive that way ;) [23:36] Yeah, I wouldn't have told you if you didn't know any better. [23:39] :P [23:41] OvenWerks: Do you think it would be useful for you to have access to that error tracker that I posted a link to? [23:42] You don't even have to be an uploader to have access to it. [23:43] I guess I do need access [23:43] I would have assumed there would be bug reports for all these things [23:44] Not necessarily. It's basically the apport logs of what happens when people don't click "report bug". [23:45] It happens more times than not. [23:45] The problem with jack errors is that someone trying to get jack to start with bogus values can lock up jack... and sometimes pulseaudio [23:45] Right, which wouldn't trigger apport. At least the python errors would trigger apport. [23:46] It is almost impossible to check for every different audio interface someone might use [23:46] a jack fail or lockup will give a python dbus error [23:49] That *might* show up, but I'm not seeing any of those on the tracker as far as I can tell. [23:51] Most of what I've seen is "UnboundLocalError: local variable 'old_pid' referenced before assignment". [23:52] Ok, that should get fixed next version [23:52] Oh good. [23:52] That's by FAR the most common error. [23:53] Either way, I still think it'd be a good idea for you to have access. You just have to request it. === tomreyn_ is now known as tomreyn