Search
Saturday, 21st of November 2020, 9:50:52 UTC
11:21:05
phoe
is it possible/feasible to add cleavir-code-utilities to quicklisp?
11:21:55
phoe
I'd like to depend on c-c-u:destructure-lambda-list for a utility of mine that needs to parse destructuring lambda lists
11:55:57
beach
I would like to make a separate code-utilities library, but it is not set up for that currently.
11:56:18
beach
Also parsing lambda lists is tricky because implementations are allowed to add new lambda-list keywords.
11:56:34
beach
scymtym is working on that problem I believe.
11:56:53
beach
Time for a long-ish break.
12:02:16
phoe
beach: I want the library to work on portable code, so only on CL lambda list keywords
12:02:41
phoe
is there any way I can help with separating code-utilities out?
12:19:23
phoe
since this library is a gathering of various necessary libraries, I propose the name Ecclesia
14:15:46
phoe
OK - how should I start? Make a new repo and try to factor this stuff out of Cleavir?
14:16:28
beach
Yes, I think so. There is a separate ASDF system and a separate package for the code utilities, so it should be straightforward.
14:16:58
phoe
I'll check if there are any missing dependencies
14:17:47
beach
You can remove acclimation if you like.
14:17:54
beach
That's the only dependency.
14:18:22
beach
And you can merge the condition reporters with the condition types.
14:18:42
beach
... or you can keep acclimation if that's what you prefer.
14:19:01
phoe
is acclimation on quicklisp?
14:21:00
phoe
OK, I'll do that and once I'm done I'll offer to transfer the repository to you
14:21:34
beach
We should put it in s-expressionists.
16:45:28
phoe
beach: initial commit is ready, https://github.com/phoe-trash/Ecclesia
16:47:52
phoe
let me know when to transfer
16:52:03
beach
I don't know how to transfer, whether you transfer to a person, to a GitHub "account", both, or neither.
16:53:06
phoe
> You don’t have the permission to create public repositories on s-expressionists
16:53:26
beach
I can give you such privilege if you tell me how.
16:53:36
phoe
it should be on https://github.com/orgs/s-expressionists/people
16:53:43
phoe
you should be able to invite a new person
16:54:03
phoe
type phoe in there and hit OK; once I join, you should be able to refresh the page
16:54:14
phoe
and then you should be able to give me the proper permissions
16:54:38
beach
Hold on. I have to remember my GitHub password.
16:54:53
phoe
yes, the hardest part of them all
16:56:51
phoe
I can create repositories there; hold on
16:57:17
phoe
ACTION runs to add it to quicklisp
21:32:18
Shinmera
scymtym: How far along was your port of beach's dispatch mechanism to SBCL?
21:34:16
scymtym
Shinmera: i only did a few experiments. the summary i wrote for you when you last asked is probably still accurate: https://techfak.de/~jmoringe/sbcl-fastgf-roadblocks.html (although SBCL might have diverged in the meantime)
21:35:16
Shinmera
Hmm, I don't remember seeing that page, but it's quite possible you showed it to me before anyway :)
21:36:39
scymtym
likely since i wrote it when you asked :P
21:36:49
Shinmera
Now that I have an actual commercial need for GFs to be Very Fast, I'm considering putting forth a bounty for it. I can't pay enough to cover the time needed to implement it on a proper salary, but should be on the order of thousands still.
21:37:01
scymtym
i think i also had notes regarding the actual experiment
21:37:59
Shinmera
Regarding the roadblocks, I don't care if it's a special branch of SBCL or needs to revalidate at startup, so they're no problem for me.
21:38:08
phoe
Shinmera: I am most curious now
21:38:38
phoe
and I'd throw some money at this issue, too
21:39:15
phoe
and likely will other people when you make a bounty of sorts for it - bountysource, or perhaps something organized by CLF
21:39:22
phoe
(to avoid fees and third party issues)
21:39:39
Shinmera
Contacting CLF could be worth it
21:40:02
Shinmera
Or could be handled through OpenCollective, I suppose.
21:40:15
phoe
sure, the tooling is not the big issue
21:41:05
Shinmera
Anyway, it's still something I'm not dead set on. Just wanted to toss it out there.
21:41:25
Shinmera
For now performance is... passable on most systems that aren't low-end.
21:41:38
Shinmera
And there's stuff I can do on my end
21:50:19
scymtym
i will have a look at the state of the project
Saturday, 21st of November 2020, 21:50:52 UTC