freenode/#lisp - IRC Chatlog
Search
6:43:44
Harag
I have a little object db I wrote, its designed in 3 logical layers basic, indexed and the kitchen sink thrown in. The idea is to use different asd files to load only what is needed for a layer. My question is should the layers be in their own package/namespace or should I keep it in the same namespace
6:49:18
Harag
that was my feeling as well but I was hesitant that I could find my self with circular references some time in the future as I add more functionality
6:59:03
Harag
I never used bknr-store, but my previous experience with clos based data stores left me disillusioned
6:59:28
PuercoPop
It can't be that naive if it has indexes. Storing things as plists seems like playing to Lisps strength more than anything.
7:08:42
Harag
:PuercoPop i call it naive because of that very fact, everything including the index is using lisp and minimal functionality, the way the data is stored in separate files often negates the need heavy indexing
7:09:12
aeth
Harag: as a last resort to avoid a circular reference, you can pass a symbol from one layer to another e.g. in a macro... it's pretty rare, though, even with one-package-per-file
7:09:15
no-defun-allowed
Okay, I broke something in SLIME and any attempt to type anything in the repl just gives me the message "Text is read-only". What should I do?
7:09:57
aeth
Harag: e.g. if A depends on B but B needs a symbol from A, then you can pass the symbol from A into B in e.g. a macro api.
7:13:15
Harag
:aeth ok, well I wont pretend I can see how that code looks in my head now but I will keep it in mind. the store will most likely only ever be used and extended by me so I will just have to solve the problem if it ever comes up
7:26:35
no-defun-allowed
doesn't like that, "Wrong type argument: integer-or-marker-p, nil" and the input has the package name font
7:27:28
ck_
hmm. Have you tried sending something to the repl already? C-c C-j or whatever the standard binding is
7:27:31
Harag
:PuercoPop ... on that point of lisp strengths, doing a reduce with string comparisons over 10k plists ... 0.004 to 0.008 seconds of real time... i can live with that
7:28:24
no-defun-allowed
C-c C-e works, I'm going to dump out a load form and restart since I made a method for that
7:29:28
ck_
Weird. Well, I have no further suggestions other than using the ielm repl to fiddle with the lisp repl buffer state
7:30:59
no-defun-allowed
I've restarted it and restored the state using make-load-form (which I wrote a method for).
7:38:35
schweers
no-defun-allowed: I’m a bit late for this, but maybe it happens again. Did you try killing the repl buffer and calling M-x slime-repl RET?
9:53:24
beach
I added "A calendar program" to this list: metamodular.com/Common-Lisp/suggested-projects.html
9:56:17
beach
A simple first version would be adequate, and more features could be added later, like sophisticated ways of handling overlapping events.
9:57:33
beach
It would be a perfect application to work on for someone who would like to use it as an opportunity to learn CLIM/McCLIM.
10:37:09
jonatack
no-defun-allowed: drive-by thought, could also do M-x sl-i-li (slime-restart-inferior-lisp)
11:00:35
no-defun-allowed
jonatack: Not really, that would lose my state, which I recovered only because I could recreate it since it's mostly transforming an online API.
11:02:29
no-defun-allowed
Yeah, killing the REPL then M-x slime-repl looks like that would cut down lossage.
12:05:30
Harag
:beach ...just curious is some one actually working on a lisp operating system or is it just an exercise on paper?
12:33:42
beach
Harag: I am. SICL is going to be a safe implementation that will be the basis of a multi-user operating system.
12:34:22
beach
Harag: My first-class global environments were designed so that several "users" can co-exist in the same system.
12:35:46
beach
Harag: It is thought that getting a bootable image is the hard part of an OS, but I don't agree. The hard part is making it safe and stable.
12:36:26
beach
In fact, the bootable part is not even terribly interesting. A lot of work could be done by running the entire thing as a Unix application.
12:37:58
beach
The hard (and interesting) parts are getting a satisfactory user environment and a satisfactory application API.
12:55:14
LdBeth
To be concise, it was aimed for language implementors so a language learner could get confused
12:58:17
beach
I also need an editable version of it for WSCL. I started trying to turn the dpANS document into a single LaTeX document (not a single file, a single document), but it is just too hard.
12:59:48
Lycurgus
LdBeth, yeah nobody can prevent that sort of thing, it's not copyright infringment to promote an asshole
13:00:13
beach
moldybits: "draft proposal for American National Standard" I believe. It's the draft of the standard, but I understand it is virtually identical to the final standard.
13:02:12
beach
Plus, it is one plain TeX document per chapter, with lots of custom macros that have been superseded by LaTeX.
13:03:05
dlowe
I made a TeX parser/macro expander, with the idea that I could manually specify macros that didn't expand.
13:04:28
dlowe
The hope is to produce something that is >90% converted and then do the rest by hand to produce the final document.
13:38:34
galdor
is there a way to get the output of the compiler when the cffi groveller fails ? I end up with a UIOP/RUN-PROGRAM:SUBPROCESS-ERROR which only contains a code, command line and process which is a UIOP/LAUNCH-PROGRAM::PROCESS-INFO with OUTPUT-STREAM set to NIL
14:23:19
galdor
it seems to end up in *inferior-lisp* indeed; still better than nothing, but I really need to write a patch so that the output end up in the condition
14:24:16
galdor
"You may not distribute the code without prior consent from me." -> well at least it is clear :)