freenode/#lisp - IRC Chatlog

3:17:47 beach Good morning everyone!

3:18:13 Josh_2 Mornin'

3:20:31 ealfonso is there a good way to destructure a REST pathname in hunchentoot, like for extracting 123 and 456 in "/a/123/b/456"

3:22:36 ealfonso it seems like query parameters are much better supported?

3:38:45 PuercoPop ealfonso: You can use a regexep and macro to wrap the boilerplate

3:39:43 PuercoPop This is what I do https://gist.github.com/PuercoPop/fb40d1cb8025272c3ba2437e71f26a3a

4:44:31 drmeister I'm learning to love LOOP

4:44:32 drmeister https://www.irccloud.com/pastebin/AWYX0oJg/

4:47:40 beach drmeister: That is a bit ugly though this all those repeated ELTs.

4:48:24 drmeister That's why I post this stuff. Like a olden times printer - putting his work up in the public square for free editing. :-)

4:48:28 drmeister What would you suggest?

4:48:52 drmeister I need to fill seven contiguous elements of a vector.

4:49:11 drmeister Four numbers for the components of a quaternion and then three x, y, z coordinates.

4:49:16 beach I don't have any immediate suggestion.

4:50:22 drmeister That function takes a vector like this:

4:50:45 drmeister #(-180.66919427590497d0 -881.2260182904439d0 231.43257852229718d0 -479.9920154012122d0 1992.9829583918222d0 -14.490829798676634d0 -67.1107262767643d0 122.77569926522422d0 840.6421670541661d0 -291.06636154816147d0 397.8898934964087d0 1971.7084478581778d0 -35.35365262319839d0 -56.41564091073567d0)

4:50:57 drmeister And turns it into this...

4:50:58 White_Flame you could use displaced arrays, or fill pointer, to hide all the ELT foo

4:50:59 drmeister https://www.irccloud.com/pastebin/IfIg3HxP/

4:51:55 White_Flame there's also REPLACE to copy a chunk in if you have it pre-formed

4:52:29 drmeister I could - but I don't have it preformed.

4:53:31 drmeister I'm taking a single long helix with a centroid and quaternion to describe the orientation and breaking it up into smaller helices that lie along the larger one.

4:53:34 phoe drmeister: (defun bar (part-position part-pos-start) (flet (((setf foo) (new-value index) (setf (elt part-position (+ part-pos-start index)) new-value))) (setf (foo 3) 42)))

4:53:59 phoe wrap this FLET around your LOOP and use the shorter SETF function

4:55:19 phoe also you can FINALLY (RETURN PART-POSITION) from inside LOOP at the end to be a little bit more idiomatic

4:56:00 White_Flame (replace part-position (list eha ehb ehc ...) :start1 part-pos-start) is probably the shortest way to do the final batch of ELTs, if I'm reading your code right

4:56:19 drmeister I'll do the finally thing - that's nicer. But I don't have a big problem with the repeated invocations of elt. They get across what I'm trying to do.

4:56:21 White_Flame and there's opportunity for the implementation to optimize that away, to avoid creating the intermediate list

4:56:21 phoe White_Flame: yes, though it'll cons

4:56:38 phoe oh, hm

4:57:15 phoe White_Flame: how does it work? I can't declare DX on looped variables

4:57:37 White_Flame if the REPLACE was unrolled

4:57:49 phoe I see

4:58:17 White_Flame or it could be stack allocated

4:58:33 White_Flame (if none of the terms need boxing)

5:01:04 drmeister I doubt Cleavir does this optimization yet - but I see your point

5:03:48 White_Flame it would be nice to use something similar to destructuring-bind to collapse the 1st set of ELTs

5:05:07 phoe I still think a FLET and DECLARE INLINE would be a bit shorter, but I agree that this current version is readable.

5:05:55 White_Flame yep

5:12:57 drmeister Not bad - it's breaking this...

5:12:58 drmeister https://i.imgur.com/GkNu6kW.png

5:13:00 drmeister Into this...

5:13:44 drmeister https://i.imgur.com/tZAaklz.png

5:14:08 drmeister But the staples (the cyan tubes) are messed up.

5:14:26 drmeister I'm too tired to fix them now and I have a midterm to give in the morning.

5:14:36 drmeister Thanks everyone - good night.

5:14:43 ecraven good night!

5:19:39 jack_rabbit How do you guys deal with providing patches for libraries? I've got a few PRs up on github, but some have been there for ~2 months with no response. Is it bad form to directly email an author? I know people are busy - am I being impatient waiting only 2 months?

5:20:09 ealfonso` PuercoPop thanks. where is the match symbol coming from? I don't see it in hunchentoot or ppcre

5:23:54 pillton jack_rabbit: I wouldn't classify you as being impatient.

5:25:25 phoe jack_rabbit: 2 months is enough for a person to take a look and respond

5:25:33 phoe I'd poke the maintainers.

5:25:55 jack_rabbit pillton, phoe, Thanks. I appreciate the input. :)

5:30:56 ealfonso` PuercoPop I ended up using something similar to yours, but using ppcre:register-groups-bind

5:32:38 jack_rabbit On a similar note, I've been looking to do more contributions to lisp projects. I'd still classify myself as a beginner. Does anyone know of any libraries or projects in need of some love?

5:34:43 mathZ I am interested in AI project(s). Are there any projects using tensorflow ?

5:36:55 pillton mathZ: Why are you asking?

5:38:11 mathZ pillton: Since i donot know such a project

5:38:25 pillton You want to write tensorflow in CL?

5:38:58 mathZ hope there is already such a wrapper

5:41:40 PuercoPop ealfonso`: Yeah sorry, I was going to comment on that. It is TRIVIA:MATCH, a pattern matcher, in this case it is basically syntactic sugar around ppcre

5:41:43 PuercoPop See https://github.com/guicho271828/trivia/blob/master/ppcre/package.lisp#L52

5:52:17 beach jack_rabbit: What kind of work are you looking for?

5:52:51 jack_rabbit beach, probably some minor work suitable for a relative CL beginner.

5:53:13 beach OK, but I suppose the purpose is to progress in your Common Lisp knowledge, right?

5:53:50 jack_rabbit I would say so, yes. Secondary would be to increase the overall quality of available CL libraries.

5:54:00 beach Sounds good.

5:54:07 beach Let me see what I can dig up.

5:54:15 jack_rabbit I can always just write my own projects if all I want to do is learn. :)

5:54:24 beach I understand.

6:00:06 beach I have an embryonic GUI application for double-entry bookkeeping, but that might not be what you are looking for.

6:00:32 beach It is a pretty simple system when it comes to logic, but there are plenty of small features that could be added to it.

6:00:40 beach And It could be made a lot prettier.

6:00:45 beach It uses McCLIM.

6:01:06 jack_rabbit Sweet! I'll take a look.

6:01:36 beach https://github.com/robert-strandh/Compta

6:02:26 beach It is also an application that I don't think anyone is using, so you can go ahead and experiment on it as much as you like.

6:02:49 jack_rabbit Thanks, beach, I appreciate your help.

6:03:17 beach Sure. I am not sure that this is what you want, of course. If not, I'll try to dig up something else.

6:04:01 beach On the other hand, it is only 450 lines of course, and more than half is the GUI.

6:04:14 beach So it is in some ways perfect for the purpose.

6:04:31 jack_rabbit Yeah, sounds like a perfect amount to be able to wrap my head around.

6:04:40 beach Definitely.

6:11:22 d4ryus1 ** NICK d4ryus

6:21:53 beach jack_rabbit: The application is also interesting because it uses CLOS a lot, including the APPEND method combination for the I/O part. So despite its small size, it exercises a lot of Common Lisp. And, it uses CLIM/McCLIM, of course, so that's positive too.

6:23:08 jack_rabbit Excellent. I've been usinc McCLIM quite a bit lately, and want to do more. I also have been looking for reasons to exercise the less common features of CLOS.

6:23:33 jack_rabbit I'll probably start playing with it tomorrow.

6:24:33 beach Great! If you have any questions at all, don't hesitate to ask.

7:23:56 muresanvlad_ ** NICK DonVlad

7:26:51 DonVlad ** NICK Murii

7:55:47 LdBeth Open Genera is so great

7:59:37 beach Yes, but we can do better.

8:11:58 sauvin_ ** NICK Sauvin