freenode/#lisp - IRC Chatlog
Search
18:17:09
shrdlu68
Haha, and I stopped inlining, which was messing profiling results. Found a major bottleck, now down to 5s with my 35kb test-file.
18:26:55
shrdlu68
shka_: sb-sprof results without inlining make much more sense: https://gist.github.com/shrdlu68/081a78c6196b676395ca518a268d9119
21:31:47
k-hos
any ideas why calling (compile-file "test.lisp" :system-p t) in ECLs repl would emit a .fasc file instead of an .o
21:54:45
k-hos
their examples on the site say :system-p t is supposed to make it emit .o files by invoking gcc
22:10:51
k-hos
but I can't find any information about how to manually specify the compiler, or get it to emit just .c files
22:34:01
oni-on-ion
k-hos: and you've seen this? https://common-lisp.net/project/ecl/static/manual/ch26.html
22:38:11
k-hos
standalone program is what I am trying to get, preferably by compiling all files involved and not loading them at run time
22:39:56
oni-on-ion
k-hos: well. bulding programs and choosing a C compiler i think are very related concepts. you did say that you checked the manual and i did not intend to not believe you but i did a google search and thats where i ended up
3:50:06
SaganMan
It's fairly good morning today. Mangkhut has passed China and things are already back to normal.
3:53:33
SaganMan
yeah, I'm now talking to friend who lives in Shenzhen. He's safe now but he had hell of experience in typhoon.
4:12:25
no-defun-allowed
(ie: if i say "same license as emacs" is it what license was used when i published the software or does it change automagically when emacs changes?)
4:14:38
sabrac
I would interpret it as what the emacs license was when you published, but I am not a copyright lawyer in any jurisdiction.
4:17:35
sabrac
I have seen license language that adds "as amended" or "as amended from time to time", which I would interpret differently.
4:24:06
mfiano
I have a DEFUN over a LET over a DEFUN, so I can initialize and also reset the state of the closure. What can do to have the call-sites to the inner function not emit compiler warnings? Is using funcall on a quoted symbol instead of #' here the only thing I can do, because that sort of feels like a hack, and ?
4:28:34
mfiano
Sure, well the code is just a pretty simple memoization closure: https://gist.github.com/mfiano/79694e884f37779b9d1e31be78f8ea77
4:29:10
mfiano
and call-sites look like (load-texture ...) but will never hit that at runtime until the toplevel initialize function is called
4:29:25
mfiano
This is only a compile time warning because the load-texture function doesn't exist then
4:29:28
loke
beach: I've always been a bit confused about the symbol FMAKUNBOUND. It's 11 characters. What system had a length limit of 11 character so as to make it impossible to use the name FMAKEUNBOUND?
4:31:23
loke
beach: Didn't some system have a limit of 6 characters? ITS Filenames certainly did. Was it originally FMAKUN?
4:33:33
beach
Or `ls', though that was from Multics. But Multics had a long name and a short name for most commands, so `ls' was short for `list'.
4:33:48
mfiano
At first thought I was thinking that'd emit a redefinition warning at runtime when I call the toplevel initialize function. I guess that is not the case
4:35:23
loke
Someone on SO mentions radix50 which packs 6 characters in one word (that's what's used for filenames in ITS). But those use 36-bit words, and Unix was for PDP-11. Also, 6 characters would fit ‘create’. :-)
5:25:12
Fare
sometimes, a hackish name is a good way to signal that the function is a hack. FMAKUNBOUND sounds like it.