freenode/#lisp - IRC Chatlog
Search
17:22:40
dim
is there a string-prefix function in the alexandria et al common libs? sounds like it's all about checking length then string<, right?
17:23:38
doudou
http://ccl.clozure.com/docs/ccl.html#programming-with-threads does it mean that we cant ask kernel to let always thread finish until i/o block it under linux? that any time kernel can stop a thread to do something?
17:24:51
minion
beach, memo from Shinmera: Regarding your advice to keep docstrings separate from code-- now that I've tried it, I have to agree with you. It's much cleaner this way. I had to write a small library to help with it a bit though: http://shinmera.github.io/documentation-utils/
17:29:07
beach
Shinmera: Looks good. However, I don't think the expressions in the body of DEFINE-DOCS are forms as the documentation and the signature claim.
17:38:48
attila_lendvai
Xach: I don't know, I'll push the fixes soon. maybe luis can make a quick release and then just retry?
17:39:25
attila_lendvai
this of course assumes that they are simple fixes in cffi, that doesn't require fixing the libs
17:40:09
attila_lendvai
Xach: I'll let you know if I don't see any more issues on my side and it's all pushed
17:51:30
max3
i know this isn't #clojure but no one is responding to me there so: if conversions is a map with functions as values (of the correct types) and vamp-key and value are both strings can someone tell me what this does: ((get conversions vamp-key) value)
17:53:05
beach
Maybe it would be better to choose the language you use according to the willingness of the IRC channel to help. :)
17:53:45
attila_lendvai
LiamH: what's the deal with gsll using :sizet? who is supposed to define it and where? (with CFFI master it's not found)
17:53:59
Bike
it grabs a function from conversions based on the key vamp-key, and calls that function on value.
17:55:24
attila_lendvai
LiamH: hrm. ok, I'll look at the diffs and see if that was a valid expectation or not. I don't remember removing it, but it may have been removed as a sideffect of some change
18:01:52
attila_lendvai
LiamH: I see now, it's this change: https://github.com/cffi/cffi/commit/4fbe5864552d6f7d8866745f371346174acea942 I have turned :sizet into a cffi::size-t. I don't think CFFI should pick random C definitions and define them in the keyword package... gsll should just refer to an internal symbol, or grovel it for itself, no? or *maybe*, cffi-libffi export it
18:04:23
LiamH
|3b|: I agree. And I agreed before, which is why I moved from GSLL (the way attila_lendvai wants) to CFFI.
18:04:26
|3b|
* has wanted it a few times, but just hard codes it based on arch since i don't feel like trying to remember how grovel works for 1 type
18:04:27
attila_lendvai
|3b|: the problem is that it's goveled, and only cffi-libffi uses the goveller
18:06:27
LiamH
attila_lendvai: If there is enough demand for other symbols, then I would say yes, include those too.
18:07:06
|3b|
yeah, if the other symbols are commonly used to define other types, or passed to functions in unrelated libs, might be reasonable to have more
18:07:14
LiamH
The idea of a library is to provide common processing. If there are enough uses for it, why make every system redefine it?
18:09:11
attila_lendvai
maybe we should open a new asdf system for the definitions of the C stdlib (as opposed to the C standard, i.e. :int and friends)
18:09:11
LiamH
attila_lendvai: What is it that iolib defines that is also defined by other systems (that don't already require iolib)?
18:10:16
LiamH
attila_lendvai: Then I agree, it would be good to have those. Maybe a separate system is the right way to go.
18:12:28
attila_lendvai
Xach: I'm afraid there won't be a fix in the next hours. maybe I could temporarily reinstate :sizet, with a big fat warning, but then a new CFFI release is needed, or ql needs to switch to master
18:12:36
LiamH
Well including :sizet was not controversial as it was in there, so please put it back so GSLL will work.
18:13:17
LiamH
attila_lendvai: You are the one who raised the idea of adding other things, so that does merit a discussion.
18:13:39
LiamH
I don't think restoring sizet merits a warning. It's been in there literally for years.
18:15:16
attila_lendvai
that's not an argument. :sizet is an arbitrary name to begin with, and that naming and the use if the keyword package itself warrants a discussion IMO...
18:15:34
attila_lendvai
I can restore it *and* write a mail, but that still needs ql to switch to master
18:16:38
attila_lendvai
size_t comes from 7.17 Common definitions <stddef.h> https://stackoverflow.com/questions/2547635/is-size-t-only-in-c-standard-or-c-standard-as-well
18:29:36
Xach_
I am using the master branch of gsll from https://gitlab.common-lisp.net/antik/gsll.git
18:35:23
attila_lendvai
I thinks it's because I was working from an image, trying with empty sbcl now
18:42:23
attila_lendvai
Xach: hrm, that's another interesting question. it's using :double as base type for the C enum, which is against the C std
18:43:59
attila_lendvai
|3b|: is there a specific reason for using double there? or could it be omitted to use the default base type?
18:50:21
attila_lendvai
hrm, I see the intention there. the c fun uses a double arg, and the valid values are consecutive #define's...
18:54:19
attila_lendvai
|3b|: what do you think about using the new cffi/c2ffi to generate a binding for opengl? and then port the lispy stuff over to that? if you feel like venturing into that I can offer to set up the c2ffi part...
19:00:07
|3b|
attila_lendvai: i think i'd rather change the function type to a 'double-or-enum' or something that coerces integers to doubles, to avoid breaking any existing code
19:00:45
|3b|
though i wouldn't mind seeing something in cffi that acts like defcenum but accepts arbitrary types
19:01:33
attila_lendvai
|3b|: I don't think anything would break if I split the defcenum into a row of defconstants, no?
19:03:06
attila_lendvai
defcenum with doubles is basically the question of whether to bring CFFI's behavior closer to the C standard, which would accommodate better for generated bindings, or not.
19:03:36
|3b|
code that passes :foo to the function would break if you named the constants anything other than :foo, and :foo is required to have the value :foo
19:04:21
attila_lendvai
|3b|: right, I didn't consider that the enum members are renamed compared to the .h and they are even keywords... scratch that idea then... :/
19:06:21
attila_lendvai
|3b|: yep, they wouldn't care in this specific case. but there are other cases where this kind of coexistence is not possible (e.g. the mismatch of the type namespaces of CFFI and C)
19:07:11
attila_lendvai
* ponders about the pro's and con's of relaxing the defcenum base type constraints
19:08:20
|3b|
(not that it gets enum types right anyway, unless someone fixed the bug i filed the other day :)
19:09:42
pareidolia
I just read that SLIME only exists since 2003. How were people doing their CL coding before then?
19:11:34
pjb
Yes, and even without ilisp, I had just a C-x C-e implemented over *inferior-lisp* that made my workflow quasi-identical to what I use of slime nowadays.
19:11:48
moore33
pareidolia: There was ilisp, which sent everything through the inferior lisp's REPL, if I recall correctly.
19:31:10
attila_lendvai
luis: Xach is testing master for now. he may need a release later for convenience...
20:00:26
jasom
hmm c-l.net gitlab won't accept ed25519 keys; I wonder if that's a gitlab or ssh issue.
1:52:14
myrkraverk
Is it possible to read data from drakma piecemeal? As in, if I want to give a status feedback on a huge download, or display a partial image.
5:04:25
beach
From looking at several recommendations (ACM, MLA, Chicago manual of style), it appears that it is not recommended to include URLs in bibliography entries for otherwise published material, and when a URL *is* included, it should be accompanied by the date of retrieval.