libera/commonlisp - IRC Chatlog

12:12:12 lisp123 Franz have some wonderful code, when reading some of their open source libraries (which by themselves are not as useful unless one is using ACL...)

12:15:25 lisp123 https://github.com/franzinc/st-json anyone use this?

12:47:50 Nilby Is there if function to get the full dependency tree [flattened or not] for an asdf system?

12:50:52 rotateq Nilby: Maybe via ql-dist:dependency-tree ?

12:51:15 Guest74 Nilby: what do you mean by 'full'?

12:53:18 Nilby Hmmm ql-dist:dependency-tree give me nil for strings symbols or asdf-systems

12:53:44 Nilby I mean every recursive dependency for every depended on system

12:54:06 Guest74 in any case asdf:system-depends-on will get you what a system depends on, I guess you could chase the rabbit from their.

12:54:11 Guest74 s/their/there/

12:55:35 Nilby Guest74: yes, but it's worse than that, because I'd have to parse :feature expressions, etc.

12:56:30 Guest74 I guess that's one thing to add to my system protocol. exhaustive-depend/encies recursive-depends/encies?

12:57:08 Guest74 why do you need to parse whatever :feature is?

12:58:31 Nilby Guest74: If you don't, then you don't know what the real dependencies are

12:59:51 Guest74 I don't understand.  what does whatever :feature is have to do with the systems a system depends on?  Is this a system that isn't loaded?

13:01:41 Nilby :feature expressions in dependencies includes different systems based on different feature sets, e.g.

13:02:55 Guest74 or, you're talking about that unintuitive keyword in an asdf defsystem.

13:04:07 Nilby yes, which is unfortunately quite essential

13:06:15 Guest74 I think that is only a problem if the system isn't loaded.

13:06:49 Guest74 or you want to know all possible dependencies?

13:07:33 mfiano Can someone remind the LOOP conformity of if binding clauses belong before or after termination clauses?

13:08:04 mfiano I _always_ forget this after many times of learning it again and again :/

13:08:10 mfiano ACTION thinks he is getting old

13:09:43 Guest74 Nilby: do you have an example system that uses this so I can test?

13:11:25 Guest74 mfiano: I can't remember either.

13:14:29 Nilby Guest74: and example is in cl+ssl.asd it depends on :sb-posix on sbcl and sb-bsd-sockets only on windows

13:19:07 Guest74 I don't see it keeping that information anywhere.  This is one of the things that bothered me about asdf

13:29:04 Nilby Ah ok, I figured out how to get it to parse the feature things: asdf/find-component:resolve-dependency-spec . Now I can have some hope of resolving the full list.

13:31:53 Guest74 let me know if you do. I'll add it to my system which I preemptively made in hopes of another build system.

13:32:25 Nilby Guest74: Sure. Thanks for helping me through this

13:32:51 Guest74 np

13:48:49 Guest74 anybody know anything about the whereabouts of LMud?  I'd be nice if I could use their boring stuff.

14:12:48 Guest74 So I found what seems to be it at https://gitlab.common-lisp.net/lmud/lmud and this nothing at all what I would expect after reading the cliki.

14:14:00 Guest74 cliki should have add some ability to tag/flag pages/systems to at least alert people and to prompt those who like to fix cliki pages.

14:14:24 boka *Emacs : Everyone should try at least in their one life*

14:15:27 Guest74 The only thing of interest in the code base is changes to some work by Arthur Lemmens which seems to be where split-sequence was conceived.

14:20:40 Guest74 err, not conceived, solidified?  whatever you call it when you finalize all the names and conventions in your protocol.

14:20:50 Nilby Here's what seems to give me the full list: https://plaster.tymoon.eu/view/3189#3189

14:21:30 Guest74 and no mention of him.  I wonder if Arthur and James Crippen weren't friends?

14:23:53 Guest74 Nilby: that's interesting, it returns a list?

14:24:02 Nilby yes

14:24:47 Nilby It probably depends on what your current *features* are

14:25:23 Guest74 I'm not seeing any lists.

14:26:04 Guest74 I've got sb-posix in my deps, but it didn't come in a list.

14:27:00 Nilby weird, I did it on a bunch of systems and it seemed to work

14:29:03 Guest74 worked as in it returned sb-bsd-sockets?

14:30:39 Nilby hmmm, you're right, there's stuff missing

14:30:58 Nilby ACTION is fixing..

14:35:44 Guest74 though a list being returned sometimes seems probable.  In my who-depends I seem to be ignoring anything that's a cons.

14:36:44 Guest74 a code comment would have really helped

14:41:33 Guest74 ok now I'm getting lists.  This makes no sense why it would change.

14:42:33 Nilby ACTION is slowly fixing

14:50:14 Guest74 Nilby: after restarting my lisp I'm not getting lists or a complete dependency listing.  It's like asdf is changing what information is provided based on some unknown criteria.

14:51:03 Nilby Hmm. That's weird. Caching ?

14:56:04 Nilby I think this fixes the bug https://plaster.tymoon.eu/view/3190#3190

14:57:49 Guest74 I'm still getting no lists for cl+ssl.  both before and after loading the system. both included sb-posix.

14:59:18 Nilby It look correct on mine, and has sb-posix. What's your (asdf:asdf-version) ?

15:00:05 Guest74 I do see them for things like dexador

15:01:04 Nilby and I tried it on CCL and I didn't get sb-posix

15:02:34 Guest74 see that's what I asked you in the beginning, if you want loaded deps or all possible deps.

15:03:52 Nilby good question, but I guess I should have said that I wanted all the deps that asdf has actually loaded

15:04:07 Nilby not that it could ever load on any system

15:05:16 Nilby I suppose the other way could be useful too

15:05:26 Guest74 I probably have an old version of cl+ssl, there is no use of :feature.

15:07:30 Shinmera here's how you properly traverse dependencies in asdf: https://plaster.tymoon.eu/view/2464

15:07:31 Guest74 Nilby: ah, it seems the reason why I was ignoring lists is that they only seem to apply to dependencies that weren't needed.

15:08:43 Nilby things like pngload, osicat, postmodern, serapeum seem to use the :feature dependencies too

15:11:18 Nilby Shinmera: Thanks!

15:11:37 Guest74 Shinmera: thanks, missed that asdf-system-defsystem-depends-on

15:21:59 Guest74 Nilby: thanks, pngload shows me that I was wrong about it only being unneeded deps in lists.

15:28:40 Guest74 so I'm guessing depends-on should have an optional resolve-dependencies that defaults to t.

15:50:45 dbotton Is there a library for working with modifying asd files for asdf?

15:53:15 Guest74 not that I know of.  I've written a basic system protocol that I add to whenever the need arises.   I mainly use it create systems and print out an appropriate asd, but it's pretty basic.

15:55:36 dbotton I want to be able when using the clog builder to add depends-on and components (or remove perhaps) based on adding various controls or when creating new forms

15:57:36 Guest74 that shouldn't be a problem.  It's basic as in it only deals with defsystem, not opcodes or anything, and nothing like  :feature in the defsystem.

15:59:23 Nilby dbotton: That's an interesting idea. One could probably make a print method that the asdf system object back to a readable form.

15:59:33 Guest74 though I'm not sure what you mean by component, like anything that qualifies as an asdf component?

15:59:37 dbotton that is the idea nilby

15:59:43 Guest74 Nilby: That's what I do.

16:00:50 dbotton I just don't know what systems are in the original asd file to make sure I out put the ones needed for that file

16:01:28 Guest74 if it was anything complicated it wouldn't work.

16:02:01 Guest74 ... hmm, unless i get the unresolved list of deps and use those to produce a more accurate asd.

16:02:38 dbotton the asdf object doesn't cover the entire defsystem in the file?

16:02:57 Guest74 I guess I should clean this up.  It sounded silly to me at the time I wrote it, but there have been a lot of questions asking about dealing with systems lately.

16:04:32 Guest74 dbotton: it appears to.  Just in my system I wasn't dealing with that since I was mainly dealing with asd at creation time and so there's stuff I never needed.

16:07:25 Guest74 dbotton: if you've got any suggestions as to what it should include I'm all ears.  I guess I'll work on getting better printing from existing systems.

16:10:10 Nilby dbotton: If it's an .asd that you control, then as long as you don't put arbitrary code in it, you should be good. Some people do put code in .asd's, but I think it's best as declarative, so you can do more easily do automation, like you suggest.

16:10:52 mfiano Code in asd files does come in very handy sometimes, since they are never compiled.

16:11:29 Guest74 how so, mfiano

16:13:39 Guest74 I guess I'd have to have a way to specify a code block and where it needs to get printed in the asd.  Unless code blocks are always after a defsystem?

16:14:27 mfiano Guest74: https://xach.livejournal.com/294639.html

16:18:28 Guest74 mfiano: I guess you're referring to the read time?  So code in the actual defsystem.

16:18:41 mfiano Also keep in mind that using asdf itself to resolve dependencies is not a good solution, since CL:LOAD can be called at runtime to bring in more dependencies as needed.

16:19:14 Guest74 but what evil person would do that???

16:19:15 mfiano No, I'm referring to load time.

16:19:26 mfiano That's when asdf files are evaluated.

16:19:32 Guest74 this is why I stuck to dealing with simple asd creation.

16:19:46 mfiano It comes in nhandy. I have done it before, and I've seen others do it.

16:21:31 Guest74 so just need a way to store and specify something like that which is system agnostic.  Not that it matters much since there's only asdf so far.

21:57:41 mfiano Hmm, my memory is failing me but this kinda looks strangely verbose, as if I'm missing something as usual, if you just want to filter based on identity: (defun foo (things x) (remove-if-not (lambda (y) (eq x y)) things))

22:01:38 mfiano I mean, the complement of this would simply be (remove x things). Seems to filter I must use a lambda expression with eq any way you slice it, which is somehow disturbing me as if I'm forgetting how to CL.

22:03:49 rotateq What about (remove x things :test #'eq) then maybe?

22:04:06 mfiano That is the complement

22:04:19 rotateq Ah hm right, okay.

22:04:53 rotateq So using (complement #'eq)? :)

22:05:05 mfiano Though a more specific/correct one given my example.

22:05:59 mfiano That is about what I'm thinking, right now, yeah...but the other half of my brain says the more verbose remove-if-not conveys the intent better.

22:06:17 mfiano I am the most indecisive person I know :)

22:06:32 rotateq Yes I like giving intentions.

22:07:10 rotateq But by now I have saved REMOVE-IF-NOT as deprecated in my head from reading clhs.

22:07:42 mfiano deprecated, to be obsolete in year ∞

22:08:50 rotateq I'm a bit confused sometimes and then think I can never achieve an advanced level.

22:09:44 rotateq I think it's still there due to hysterical raisins, like the :test-not keyword in some functions.

22:10:05 mfiano Perhaps you aren't working on a large enough project that interests you enough to spend time gaining experience in the software architecture and algorithmic problems CL can be malleable to solve.

22:10:56 rotateq :(

22:11:37 mfiano Find something that interests you deeply, and write up a quick prototype, and iterate on that.

22:12:36 rotateq I really don't lack interests, or rather such areas.

22:12:53 rotateq Semantically a REMOVE-IF-NOT is like a filter, right?

22:12:55 mfiano You can't gain wisdom from intelligence, only experience.

22:13:02 mfiano Basically, stop reading and start writing.

22:13:20 mfiano Yes it is

22:13:25 rotateq I know, and I have to compensate with experience.

22:14:47 mfiano Compensate? Why?

22:15:38 rotateq Not so important now. :)

22:15:46 mfiano You lost me, but ok.

22:16:36 rotateq No problem, do you now have a good idea for what you asked?

22:18:13 mfiano Yeah, remove-if-not and stop guessing my memory. Much code must be written with finite time.

22:20:20 rotateq Good.