| 24 Jan 2026 | |||
| timo | :D | 19:58 | |
| a wild yak has appeared | 19:59 | ||
| japhb | If it's a fully owned subsidiary, it isn't EU sovereign, kinda by definition | 20:01 | |
| (But yeah, I did hear about that. :eyeroll:) | 20:02 | ||
| timo | shhh don't tell anybody! | ||
| japhb | Wild yak! Domesticate it now! | ||
| timo | (the error above just happens because `cro services` looks for any .cro.yml below the current directory, and my workdir was still / in this container) | ||
| typical case of DIHWIDT | |||
| japhb | heh | 20:03 | |
| timo | i'm not sure where to find the absolute best state-of-the-art build-docker-with-github-actions stuff can be cargo-culted from | 20:21 | |
| japhb | That's a good question. I have no idea. I was thinking "What about Docker's own official builds?" but I don't know if it's possible to see the machinery there. | 20:30 | |
| timo | there's a "generate jobs" workflow that spits out a matrix for the right builds like this: github.com/docker-library/official...1361079127 | 20:40 | |
| github.com/docker-library/official...rkflow#L35 - this is where the generate jobs job lives and it uses a shell script from ... | |||
| github.com/docker-library/official...enerate.sh | 20:41 | ||
| maybe it's simpler from the perspective of github actions to give every repo its own "build and publish a docker image" workflow | 20:59 | ||
| i don't think i have the brainspace for this task today | 21:02 | ||
| japhb | If feel you on that count ... | 21:06 | |
| My brain is making that "starter motor insufficient" sound right now | |||
| 7 Feb 2026 | |||
| SmokeMachine | aruniecrisps, librasteve: should we discuss that here? | 20:30 | |
| librasteve | good idea | 21:14 | |
| @arun - are you able to see this? | |||
| aruniecrisps | @SmokeMachine @librasteve i can see your messages | 21:18 | |
| SmokeMachine | Would it make sense to make a plan how and what we should achieve? | 21:20 | |
| librasteve | yep - can you give us a starter | 21:45 | |
| (actually may be better to to do this as a Cro issue so that we can have a more permanent thread) | 21:46 | ||
| aruniecrisps | @librasteve are you asking me for the starter or @SmokeMachine | 23:37 | |
| 9 Feb 2026 | |||
| librasteve | I don't mind who makes the first stab at this ... but I expect that only SmokeMachine can tell us that named endpoints will help to avoid the MetaCromponent gnarly EVAL code and that only he can specify the reqts for that in Cro that will handle all of the is accessible trait features (eg HTTP actions, user auth and so on). Is this "clear" to you? | 21:41 | |
| Sorry I don't want to assume that you do not know this stuff ... | |||
| 11 Feb 2026 | |||
| aruniecrisps | @SmokeMachine do you have any thoughts on what the first step should be? | 20:09 | |
| SmokeMachine | I think I'd say name routes | 20:10 | |
| aruniecrisps | Right, what would be the best way of approaching named routes in your opinion? | 20:23 | |
| SmokeMachine | i just remembered this exists: github.com/croservices/cro-http/pull/208 | 20:52 | |
| but originally, before I knew that PR existed, my plan was to do something this: github.com/croservices/cro-http/pull/207 | 20:54 | ||
| this test file shows what I had in mind: github.com/croservices/cro-http/pu...f2c1a9c902 | 20:56 | ||
| 12 Feb 2026 | |||
| librasteve | .oO - i will try and take a look today | 11:20 | |
| arunvickram | I will likely not be able to do look at this today, I'll probably be grinding LeetCode | 16:17 | |
| SmokeMachine: i do wonder though how this will alleviate the EVAL calls, it looks like you're defining named routes similarly to how unnamed routes are defined: github.com/croservices/cro-http/pu...f2c1a9c902 | 16:19 | ||
| 16 Feb 2026 | |||
| coleman | does Cro support SSE? github.com/librasteve/Air/issues/4...3909579354 | 17:02 | |
| patrickb | I've. just quickly read up on them. | 17:10 | |
| IIUC, it's a *client side* API on top of a plain XHR that is kept open. | 17:11 | ||
| So on the server side, it's just a normal request that the server is supposed to simply not close, but keeps pushing data into. | |||
| If that is true, then (without trying) I think Cro already allows to do that. | 17:16 | ||
| Simply set the response body to a supply and then emit blobs into it. | |||
| See cro.raku.org/docs/reference/cro-http-message at the very bottom. | 17:18 | ||
| I didn't look into how the data needs to look that is sent via this non closing response, but that's probably a trivial issue. | 17:20 | ||
| nemokosch | I wonder how this website is hosted | 17:27 | |
| arunvickram | coleman: it does support sending text/event-stream responses back to the client, my DataStar library basically generates a text/event-stream response | 17:49 | |
| i haven't tested SSE with channels/supplies yet but to my understanding it does work | 17:50 | ||
| coleman | thanks folks | 18:00 | |
| patrickb | just found raku-advent.blog/2022/12/08/day-8-...now-later/ | 18:08 | |
| 17 Feb 2026 | |||
| librasteve | .oO | 08:25 | |
| 19 Feb 2026 | |||
| Geth | cro-http: bduggan++ created pull request #211: Fix content-length in http/2 |
15:54 | |