Sender | Message | Time |
---|---|---|
8 Dec 2024 | ||
KFears (annoying) | In reply to @irenes:matrix.orgThis is something that I wanted (but failed) to say. The mindset of "just one more refactor" is dangerous: people wanting to contribute are blocked, and people contributing are burning out because they are doing everything (and too much, probably) themselves | 12:02:35 |
KFears (annoying) | I appreciate the very deep thought that goes into infra choices, tooling, integration work, releng, and more; but it feels like those things burn out people really hard and can be offloaded to other people more | 12:05:58 |
piegames | Yeah but the flipside is: If you want to contribute now, you'll be doing refactoring as well. I think it is important to be upfront about that | 12:09:16 |
KFears (annoying) | In reply to @kfears:matrix.orgThere's no harm in asking for help; sure, it might take some occasional investment to answer questions, but little things like that can compound into great results | 12:18:48 |
KFears (annoying) | Asking for CL reviews for C++ code is already in a good state! Now we "just" need to start asking for help on infra, installer, releng and other stuff :) | 12:20:13 |
ToxicFrog | This reminds me that I still need to clean up and land my RISC-V patch for the installer | 15:46:26 |
shawn8901 set a profile picture. | 19:21:14 | |
9 Dec 2024 | ||
piegames | jade_: I've heard you want to cut a release Very Soon, what's your opinion on including vs not including https://gerrit.lix.systems/c/lix/+/2190 in the release? It's a mildly breaking change, do we want to get it out early to gather feedback or do we want to have it on main for a while in the hope that the beta testers will catch any issues until the next release? | 10:32:27 |
V. 🏳️⚧️ | What is the difference between
| 12:04:14 |
piegames | V. 🏳️⚧️: Unfortunately, probably only way to find out is to look at the implementations. Per convention as I know it, one is a pure operation on the path, and the other one queries the file system. But then why canonPath has a resolveSymlinks flag and what it does differently to realPath in that case is beyond me | 13:35:09 |
V. 🏳️⚧️ | canonPath has a manual limit of 1024 followed symlinks | 13:58:41 |
V. 🏳️⚧️ | It's not needed though, ELOOP is already a thing | 13:59:27 |
V. 🏳️⚧️ | * It's not needed though, realpath considers "too many symlinks" an error anyway | 14:00:59 |
9999years | jade and i are about to be more active -- jade is starting a job at my employer and we're going to be spending some portion of our time on lix | 17:58:34 |
9999years | it's definitely at least partially a flywheel effect thing -- i would spend more time working on lix, but cl review bandwidth is quite limited, and i'm not sure where to start with some of the bigger projects (adding telemetry, optimizing the evaluator, lazy trees) | 17:59:36 |
piegames | Horrors and I are already working quite a bit on the evaluator, but rewriting the GC is still open :) | 18:39:31 |
9999years | In reply to @piegames:flausch.sociali'll want to collaborate on that more soon, but if you have any writeups of what you're working on so far that would be awesome | 18:45:48 |
9999years | i remember horrors had a bytecode evaluator draft at one point, i'm very interested in seeing that and/or reviving it...? | 18:46:05 |
piegames | Horrors is currently splitting EvalState and working on asyncification | 18:46:32 |
9999years | btw jade_ i finally uploaded the lix release process notes like you asked. i had to do battle with git hooks for this (git hooks do not play very nicely with worktrees apparently, but pre-commit fixes it automatically, at least sometimes...) https://gerrit.lix.systems/c/lix/+/2294 | 18:46:46 |
piegames | I have a mini-bytecode thingie running locally which is vaguely based on Horrors' patch | 18:47:08 |
jade_ | In reply to @piegames:flausch.sociali am unsure who uses that API if anyone. it might be better to try to cut a release first to give the explosions a little time to show up. i would like to do it in the next week if i can. | 21:20:47 |
jade_ | In reply to @piegames:flausch.socialagreed. it could be async python for all it matters | 21:29:01 |
jade_ | i have a suspicion that a lot of parts of pytest could be reused | 21:29:54 |
jade_ | for example, the nice assertions UX. we would just have to toss their main loop basically i think? | 21:30:10 |
jade_ | In reply to @raitobezarius:matrix.org i happen to be in seattle at the moment and i might do some irl work with wiggles on trying to get releng into a better state. my view of my role is to try to shepherd acceptable code in as much as possible and try to reduce dependencies on myself and core. one related thing is that we have, for basically accidental reasons, not simply given away commit access to more competent qualified people, which is 100% a governance problem of not being quorate enough due to inactivity but also that i think we just need to simply be giving it away harder with less specific requirements. there is still also the open question of what the difference should be between core and commit access (since core is already too many people to be easily quorate and obviously needs governance reform, but this takes such immense energy that i dont know if people have it) | 21:36:00 |
raitobezarius | i mean, i believe we have to do the governance reform at some point and power through it rather than powering through anything, but there needs to be a quorum among active core members to do it first | 21:37:40 |
jade_ | there is significant short-term chaos in the lives of several members of the core team for personal reasons that should become a bit less chaotic in a month or two, which should help with being able to do things. i do agree that our communication is very dysfunctional and our planning is anything but transparent and well written down. this is, sadly, a lot of work to fix, but it does need to get better | 21:44:30 |
jade_ | (mostly the lack of transparency is actually it not being written down. we do not have secret plans, just poorly documented ones) | 21:44:52 |
Irenes | reviewnig CLs is one of the | 22:39:16 |