| 18 Apr 2024 |
 fleimgruber (fleimgruber) | Yeah, for now there is a layer of multiprocessing on top. I wanted to use my problem as an excuse to look into trio again :) | 14:13:34 |
| fleimgruber (fleimgruber) | * Yeah, for now there is a layer of multiprocessing on top that calls trio_sync_top_level(). I wanted to use my problem as an excuse to look into trio again :) | 14:13:54 |
 mikenerone | Cool, just wanted to point it out. | 14:14:55 |
| fleimgruber (fleimgruber) | It has been a while since I used threading, but I quite like the structured way of thinking and trio API. If the 3rd party lib becomes async at some point, the changes will be minimal to use it that way directly. | 14:15:58 |
| fleimgruber (fleimgruber) | Another comment: I found my way to the solution via interaction here. I found a nice description and solution to my problem in the docs here: https://trio.readthedocs.io/en/stable/reference-core.html#threads-if-you-must | 14:17:24 |
| fleimgruber (fleimgruber) | I can see why this was put under the "Threads (if you must)" section, but that is already more concerned with the solution and mentions the problem (trio with blocking non-async 3rd party libs) in passing. Is there an FAQ or something where this problem could be (or was) put more prominently? | 14:19:48 |
| fleimgruber (fleimgruber) | * I can see why this was put under the "Threads (if you must)" section, but that is already more concerned with the solution and mentions the problem (trio with blocking non-async 3rd party libs) only in passing. Is there an FAQ or something where this problem could be (or was) put more prominently? | 14:20:11 |
| mikenerone | This seems like the spot for it. TBH, it seems clear to me as it is, but I'm weird. :P I'm sure folks here are open to a docs PR if it clarifies. | 14:25:55 |
| mikenerone | Driving me kinda nuts when Element (or maybe Matrix) occasionally marks a bunch of very old threads as unread. | 14:29:11 |
| fleimgruber (fleimgruber) | Yeah, I totally get why it's there, not saying to change that, just mention the problem in a more high level, e.g. FAQ, kind of place for new users that might not have all the concepts internalized enough to know by themselves that threads would be the solution. | 14:31:07 |
| fleimgruber (fleimgruber) | * Yeah, I totally get why it's there, not saying to change that, just mention the problem in a more high level, e.g. FAQ, kind of place as well for new users that might not have all the concepts internalized enough to know by themselves that threads would be the solution. | 14:31:20 |
 richardsheridan | it's a bit odd trio lacks an FAQ at all. at inception of course there were none but now we know "how do i use my sync lib" and "how do i get return values from tasks" should be in there. | 14:33:30 |
| mikenerone | oic - this is easy to find if you already have the word "thread" in your head, but not if you don't. We don't really have a FAQ page that I'm aware of. All I can think of it maybe a quick mention somewhere in the tutorial, since it's something that beginners are probably reading. | 14:33:51 |
| fleimgruber (fleimgruber) | I was just guessing that threads might be a worthwile avenue in https://matrix.to/#/!OqDVTrmPstKzivLwZW:gitter.im/$ey6qmRcBILETetArc_EZGmOLBDx2RSE9_Y9R22KtiBk?via=gitter.im&via=matrix.org&via=ancap.tech | 14:34:12 |
| mikenerone | Good news, everybody, we have a volunteer to write a FAQ for us. :D | 14:36:36 |
| mikenerone | * oic - this is easy to find if you already have the word "thread" in your head, but not if you don't. We don't really have a FAQ page that I'm aware of. All I can think of is maybe a quick mention somewhere in the tutorial, since it's something that beginners are probably reading. | 18:24:01 |
 agronholm (Alex Grönholm) | jakkdl (John Litborn): dropped a new release of exceptiongroup | 21:25:37 |
| 19 Apr 2024 |
 oremanj (Joshua Oreman) | trio-asyncio v0.14.1 released to PyPI with a bugfix for deadlock when many tasks call run_in_executor simultaneously + test updates to cope with strict exception groups | 03:56:22 |
| 23 Apr 2024 |
| oremanj (Joshua Oreman) | trio-asyncio async generator support would appreciate review: https://github.com/python-trio/trio-asyncio/pull/150 | 06:45:44 |
| 24 Apr 2024 |
| mikenerone | Is there an easy way to make an exception group trace display further than the default max_group_depth of 10? The TracebackException object isn't created by my code, so I'm not seeing how to influence that parameter. | 13:19:30 |
| oremanj (Joshua Oreman) | trio-asyncio v0.15.0 released with async generator support and a bugfix for code that calls sys.exit() within asyncio.run(). | 22:25:38 |
| 25 Apr 2024 |
|  @primordialqueef:matrix.org joined the room. | 03:40:39 |
| @primordialqueef:matrix.org joined the room. | 03:48:35 |
| @primordialqueef:matrix.org left the room. | 03:48:58 |
 jakkdl (John Litborn) | How are people currently silencing TrioDeprecationWarning if at all? It can be removed now, with trio raising DeprecationWarning like everybody else. Might be nice to save people the hassle of updating their warning filters though: https://github.com/python-trio/trio/issues/2992 | 12:31:20 |
| mikenerone | oremanj (Joshua Oreman): Do TreeVars capture the context at the time a nursery is created, such that TreeVars set within the nursery block won't have those new values seen by new child tasks? | 18:57:48 |
| oremanj (Joshua Oreman) | That's correct | 18:58:26 |
| mikenerone | Ok, that confirms what I saw then. :P | 18:58:57 |
| mikenerone | Thanks | 18:58:59 |
| oremanj (Joshua Oreman) | The model is that if you write a context manager async with my_special_scope(): that sets a TreeVar on entry and unsets it on exit, then you will see that setting only if you are lexically within the context. | 18:59:44 |
