<sneek>Welcome back nalaginrut, you have 1 message. <sneek>nalaginrut, ArneBab says: yay on a blog post about FFI! <nalaginrut>is it proper if I set "pcre_free" pointer to a pcre object pointer? <nalaginrut>I didn't free it explicitly since it always crash while freeing <mark_weaver>what do you mean by "set 'pcre_free' pointer to a pcre object pointer?" ? <nalaginrut>mark_weaver: yes irregex works, but I've tried FFI to bind libpcre <nalaginrut>mark_weaver: now I think it's reasonable to free the pcre object returned by pcre_compile <nalaginrut>and I think irregex doesn't support full PCRE according to the manual, what's strange is that it claims PCRE has full support in the end, which sounds do contradict ***michel_mno_afk is now known as michel_mno
<ArneBab_>When given a symbol in a list I want to eval later, can I add source-properties? (supports-source-properties? obj) only answers yes for very few expressions (like strings and quoted lists), but (set-source-property obj prop value) works for many more. <ArneBab_>For some symbols I get “Wrong type argument in position 1 (expecting non-immediate)” <taylanub>ArneBab_: perhaps you can turn them into syntax early on? <ArneBab_>I call read on ports which already seeked to the symbol <taylanub>never mind, I guess that doesn't work if you aren't evaling anything yet <taylanub>maybe you can compile the data incrementally or so <ArneBab_>I fear that might give trouble because I still need to restructure parts of it (replacing temporary symbols and such) ***michel_mno is now known as michel_mno_afk
***michel_mno_afk is now known as michel_mno
***michel_mno is now known as michel_mno_afk
***michel_mno_afk is now known as michel_mno
<b4283>is there a cross-platform variable for newline ? <b4283>there's (newline) but it's only for ports, and i'm looking for a string <artyom-poptsov>b4283: Hi. Probably #\\newline character is what you're looking for. <b4283>but i wonder does it expand to CRLF under windows... <taylanub>nope, it's explicitly the LF character. a cross-platform newline could be made like (with-output-to-string (newline)) <taylanub>ASSUMING our newline procedure does the right thing under Windows <taylanub>np. (and dunno if there's perhaps a shorter way) <b4283>(with-output-to-string (lambda () (newline))) <b4283>taylanub: that IS short ! ha, thanks again <taylanub>b4283: hm, I'm reading libguile/print.c where newline is defined, and I think it only outputs \\n :\\ <b4283>i'll just define it as #\\newline then, deal with it later <taylanub>#\\newline is a character constant. you could bind it to some variable. <wingo>#\\newline is not a platform-independant thing <wingo>it is one specific unicode codepoint <wingo>if you have a port open in text mode i think #\\newline generates CR+LF on windows; not sure tho ***dje is now known as xdje
<dsmith-work>wingo: Just curious: Any more major changes/improvements for 2.2? <wingo>dsmith-work: dunno :) i might redo the intermediate language; we might get unboxed float arithmetic; who knows what could happen if we incorporate libgc... <wingo>but we need to get the train rolling <wingo>but, i wanted to give a crack to the IL because maybe it would make bootstrap less terrible <dsmith-work>wingo: Wasn't the unboxing only useful on 64bit machines? <davexunit>I had a good chat about Scheme with Geoffrey Knauth, the FSF's treasurer, at LibrePlanet. <davexunit>I didn't realize who he was at first, I just noticed that he had a Racket sticker on his laptop. <mark_weaver>I've only spoken to him once, about 14 years ago or so <b4283>i'm trying to use characters as assoc-list keys, and assq-ref to retrieve them <mark_weaver>b4283: technically, eq? is not safe to use on characters <b4283>but i'm not sure if this is ok since assq-ref uses eq? to test them <b4283>mark_weaver: yeah, that was my question <mark_weaver>in practice, it currently works on Guile and most Schemes, but you should use assv-ref instead. ***michel_mno is now known as michel_mno_afk
<daviid>is there a way to ask guild compile to first import a module? <mark_weaver>daviid: apparently not. what's your use case? why not just add a 'use-modules' form to the file you're compiling? <daviid>mark_weaver: i will try something else, tx <ArneBab_>mark_weaver: I managed to preserve parts of the source-properties in the wisp-parsing, but I’m stuck at making it work for everything. Could you have a look? <ArneBab_>the main parts are in (wisp-propagate-source-properties code) and (wisp-add-source-properties-from source target)