freenode/#lisp - IRC Chatlog

22:40:32 christiansen ** NICK christiansen`

23:43:34 Posterdati hi

23:43:55 Posterdati anyone does solve the problems with antik+gsl and ecl?

23:52:49 akkad Posterdati:got any errors?

23:53:22 Posterdati sure, the problem is related the the readtable...

0:20:24 vtomole I'm having trouble running ansi-test when i run (load "doit.lsp") I get "Pathname components from SOURCE and FROM args to TRANSLATE-PATHNAME did not match: :NEWEST NIL"

0:20:40 vtomole What should I do?

1:02:57 JuanDaugherty pastebin doit.lsp

1:03:59 JuanDaugherty #clnoobs may help too

1:13:24 vtomole https://ipfs.io/ipfs/QmU9AePusNu6M9aMKvWmXVHmQpuSFqs6U5UJBdTbcLqwFQ

1:50:24 vtomole #clnoobs seems to be empty right now....

3:29:05 warweasle Are there any "manual garbage collection" systems for lisp?

3:31:58 loke warweasle: What does that mean? There is a gc function...

3:32:30 loke (sb-ext:gc) to be precise.

3:33:28 warweasle loke: GC doesn't happen right away on sbcl. I have some objects who's finalizers dont get called for a long time after they're gone.

3:33:48 warweasle I've tried cleaning out swank's data too.

3:34:17 loke warweasle: You shouldn't rely on the finalisers being called immediately.

3:34:20 loke It's just like Java.

3:34:36 warweasle loke: Right..So I have to do this manually.

3:34:37 loke What are you reying to do in the finaliser?

3:34:55 warweasle loke: Destroy opengl objects.

3:35:23 warweasle It sends an event to delete the object.

3:36:21 pillton I'd argue that you should be using the with-opengl-object paradigm.

3:36:30 loke warweasle: I would have a (with-XXX ...) macro that handles that. In addition to that, you can use the finaliser as a last-resort kinda thing.

3:36:47 pillton An implementation is allowed to not collect garbage from what I remember.

3:37:10 loke pillton: Yes, and besides, the finalisation mechanism isn't part of the spec.

3:37:22 warweasle loke: That won't work. I have to keep these things around in an event driven archetecture.

3:37:33 pillton What?

3:37:46 |3b| pillton: with-foo doesn't really work well when your scopes aren't really nested

3:37:54 loke warweasle: Well, then you have to provide a release-XXX function. Like I said, you can still keep the finalisers.

3:38:00 warweasle |3b|: Thanks. I couln't think of the words.

3:38:24 warweasle Ok, well it doesn't sound like there is a pre-existing library then.

3:38:41 |3b| * 's alternative is with-gl-resource-manager or something, and scopes that work with that

3:38:51 blob ** NICK k-os

3:38:53 |3b| * hasn't actually figured out how it will work yet though :(

3:38:59 pillton warweasle: Streams provide a close operation.

3:39:33 warweasle |3b|: Me too. I want to combine a topological/logic based system for both the render queue and garbage collection.

3:39:34 pillton I would implement a finalizer to call "close" and also warn that you haven't called close.

3:40:04 |3b| pillton: figuring out the best way to handle the 'close' call is the problem

3:40:33 loke What pillton said.

3:40:50 warweasle |3b|: I use SDL's in-main-thread.

3:40:55 |3b| (but finalizer telling you when you messed up is still useful)

3:41:11 |3b| warweasle: i mean figuring out when to call it, not how

3:42:01 pillton Are opengl resources shared?

3:42:16 |3b| which definition of "shared"?

3:42:42 |3b| (or shared between what?)

3:42:50 warweasle pillton: They are ID's which are unique. Also, opengl calls can only be called by a single thread.

3:43:04 pillton Well, an event driven implementation typically has a controller which receives events and distributes them to one or more views.

3:43:29 |3b| thats a bit of an oversimplification, you can call them from multiple threads, it just usually isn't worth the extra complication

3:45:22 |3b| there might be multiple views showing the same objects (or single objects used multiple times in a given scene). objects might share textures

3:46:12 |3b| in the general case, objects might be created and destroyed arbitrarily with overlapping lifetimes

3:46:23 pillton Ok.

3:46:34 |3b| (objects in the 3d sense rather than programming in both of those)

3:49:30 pillton What is wrong with having a gl-resource-manager with "open" and "close" operations? It seems sensible.

3:50:05 |3b| that was my suggestion

3:51:03 |3b| though it usually ends up more complicated than it sounds, since you want to avoid destroying then immediately recreating things when switching levels or whatever, so needs some sort of 'transactions' or whatever

3:52:30 |3b| and also want to avoid having union of both sets of objects loaded at once

3:54:28 pillton Can you rephrase the last sentence?

4:10:16 |3b| pillton: about union of both sets? i mean you don't want to keep the old data around until you finish loading the new data, since having both loaded at once might take too much space

4:12:59 pillton |3b|: Yes. Thanks.

6:55:32 burtons You all might find the title(s) to my new album relevant: http://soundcloud.com/burton-samograd/sets/my-kingdom-for-a-macro

6:56:50 loke`` burtons: Interesting stuff.

6:57:00 burtons thanks

6:57:09 burtons i call it 'industrial ambient noise'

6:57:15 loke`` I have to listen to all of it first.

6:57:25 loke`` The two of my favourite genres :-)

6:57:33 burtons cool

6:57:34 loke`` Three i mean

6:57:43 burtons i'm pretty proud of it

6:57:49 loke`` burtons: You a fan of Pan Sonic?

6:57:53 burtons it just kinda came out pretty fast when i started writing it the other day

6:58:01 burtons no, I can't say I've heard of them

6:58:22 loke`` They used to be called Panasonic

6:58:29 burtons haha

6:58:34 burtons clever renaming

6:59:09 loke`` burtons: If you're on Spotify, https://open.spotify.com/album/1LWY3cdCXjtwWnGfUVUSuX

6:59:21 burtons i'm not

6:59:33 burtons i should have an album on spotify i think

6:59:38 burtons something different, mellower

6:59:48 burtons called 'listener'

7:00:27 loke`` burtons: This is one of my favuorites by them: https://www.youtube.com/watch?v=4iPvxWnbOCk

7:01:34 loke`` The story goes, as I was told, that when they did their first live gig, they didn't have a name, but the event was sponsored by Panasonic so there was this big sponsor banner behind them. Thus, everybody thought they were called Panasonic and they just went with it.

7:01:48 loke`` Afterwards, the company caught on and they had to change, so they made it Pan_sonic.

7:03:22 burtons nice story

7:03:29 burtons nice track too, thanks for sharing

7:04:41 loke`` burtons: At least for their older stuff, they were using a homemade sythesizer :-)

7:04:57 burtons i sometimes do stuff like that with Pure Data

7:05:13 loke`` What is that?

7:06:08 burtons Open source version of Max/MSP, which is a visual programming environment for working with DSP streams

7:06:12 burtons https://puredata.info/

7:06:32 burtons And a bunch more stuff, you can make your whole synth using it

7:07:23 burtons here's one

7:07:25 burtons https://soundcloud.com/kruhft/untitled-4

7:08:00 burtons the image for the song is the synth 'program'

7:10:24 loke`` Listening.

7:10:46 burtons there's a few more, of varying quality

8:36:01 flip214 clhs format

8:36:01 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_format.htm

8:37:55 jurov where are actual format strings specified? i fail to find them from ^ there

8:38:29 jurov or they are not part of CL spec?

8:40:01 jurov ahh i'm blind, it's the section 22.3 link

9:06:53 flip214 when using clon to parse a command line, how would I get the non-option data?

9:07:18 flip214 Eg. for something like "git diff --options..." I'd need to first fetch "diff" to know which switches to allow