libera/#lisp - IRC Chatlog

12:26:24 ak-coram` jackdaniel: it was quite trivial in the end, I needed to set linker flags and also had some unused symbols which really didn't exist :)

12:26:34 ak-coram` thanks again for the help!

12:30:15 JohnMS_WORK Is there a lisp equivalent to var++ or var+=1?

12:30:54 ak-coram` JohnMS_WORK: see incf and decf

12:31:17 JohnMS_WORK Thanks my web searches were all showing setq. :)

12:31:59 ak-coram` np

12:32:11 jackdaniel ak-coram`: cool, I'm glad that I could help

12:40:19 ak-coram` jackdaniel: I was wondering whether using :c/c++ eliminates the need for libffi

12:40:36 jackdaniel it does

12:40:39 ak-coram` the library I'm wrapping normally relies on it, since I need to pass structs by value

12:40:44 jackdaniel but then you must have the c compiler available

12:41:06 jackdaniel (ecl may be built with only the bytecodes compiler)

12:41:50 jackdaniel ideally if there was no problem with "extern ..." declarations I've mentioned ealier, c/c++ would be a default when ecl has c compiler and dffi when it doesn't have it

12:42:41 ak-coram` right, I can see how that would be beneficial

12:52:41 ak-coram` compiling ironclad takes quite some time, it's a shame uuid relies on the whole thing when it really only needs two digest functions (md5 / sha1)

12:53:52 jackdaniel I think that there was a separate uuid library that did not rely on ironclad

12:53:55 jackdaniel but I may remember wrong

12:54:21 ak-coram` jackdaniel: there's a workaround in it for clasp which provides these digests

13:04:01 ak-coram` jackdaniel: maybe this one? https://github.com/deterministic-arts/DartsCLUUID

13:04:30 ak-coram` seems this is ironclad too

13:04:40 jackdaniel I might have remembered wrong

13:05:03 ak-coram` there's also something called random-uuid in ql, but I can't find it anywhere else

13:05:25 ak-coram` https://git.mfiano.net/mfiano/random-uuid

13:05:35 ak-coram` looks like this one is gone

13:57:20 krl_ ** NICK krl

14:41:57 puke ** NICK Guest2959

14:41:58 pyooq ** NICK puke

15:14:27 splittist ak-coram`: do you need a 'real' UUID? Or would (random (1- (expt 2 128))) (or whatever) work?

17:25:45 ak-coram` splittist: currently I use it to represent the database UUID type in cl-duckdb

17:26:43 ak-coram` might as well switch to strings, but since I get the UUID in binary form, it seemed nice and efficient to directly create these

17:27:10 ak-coram` https://github.com/ak-coram/cl-duckdb/blob/main/duckdb-api.lisp#L77

17:30:54 ak-coram` but I guess it depends on what people use this for

17:31:15 ak-coram` if you're already using the uuid package, this works great

17:31:51 ak-coram` I'm also just a bit annoyed by my 30m ECL CI runs...

17:46:07 jackdaniel most of the compilation time is spend in gcc, so it is not something I can easily address :C

23:20:42 ak-coram` jackdaniel: no blame there :)

23:20:58 ak-coram` also I have no issues with ironclad

23:21:10 ak-coram` just maybe uuid could be done without it

23:21:41 ak-coram` as evidenced by clasp