IRC channel logs


back to list of logs

***fangism is now known as fanigsm-nomnom
***wigs` is now known as wigs
<wigs>so the character of rnrs and srfi seems to be shifting
<wigs>or is it just me?
<mark_weaver>wigs: what do you mean?
<mark_weaver>not that I necessarily disagree, just curious what shift you're seeing in particular.
<wigs>leaning towards overspecifying
<wigs>the "large" language, etc. when there is already srfi
<mark_weaver>R6RS certainly had that tendency. I'm not sure I see the same thing in R6RS (although I see other problems there).
<wigs>and damnit, we dont need 'make-foo-accessor' to be specified since there is already '(cut foo-ref x <>)'
<mark_weaver>that second "R6RS" should have been "R7RS"
<wigs>I see the appeal of the unification goal, but it seems to be removing the simple, elegant spec. that used to be
<wigs>I never finished reading r6rs
<mark_weaver>I can sympathize with what you're saying, but at this point I've come to appreciate a number of different perspectives and now I'm not sure what to think. The Scheme community has such a diversity of constituencies pulling it in different directions. At this point it's hard to find agreement on how the language should be improved and (ideally) simplified.
<mark_weaver>Of course, the simplicity of Scheme is a lot of what made it appealing to me in the first place, but on the other hand it would be nice to be able to write a larger set of programs portably.
<wigs>simplification is key
<mark_weaver>Along another axis, I can sympathize with some of the ways R6RS tried to enable more efficient compilation, but on the other hand I appreciate a fully dynamic system where anything can be rebound at any time.
<wigs>implementation-specific flags
<mark_weaver>Regarding simplification, I'd probably be tempted to do things like (A) ditch the module and record systems altogether, and implement both of them in terms of normal procedures. (B) get rid of multiple values, and make pairs immutable so that the compiler could do the same kinds of optimizations on returned lists.
<wigs>(for compilation hacks)
<wigs>right, r6rs records are not pretty, and I have not yet considered r7rs offering
<mark_weaver>I can appreciate the value of a fairly large standard library, but I agree that the language core should be as simple as possible.
<wigs>are literal pairs are already supposed to be immutable?
<mark_weaver>yes. all literals are immutable, regardless of their type.
<mark_weaver>guile does not currently enforce this for most things, unfortunately.
<nalaginrut>morning guilers~
<mark_weaver>but in practice that doesn't help much, because the compiler can rarely know whether a given variable will always hold a literal.
<mark_weaver>hi nalaginrut!
<nalaginrut>mark_weaver: hi mark! long time no see ya
<mark_weaver>Indeed, I've been away from irc for a while. How are you?
<nalaginrut>mark_weaver: well, that's fine, just no direction what should I do, my TODO is too messy
<taylanub>wigs: R7RS records are just SRFI-9.
<taylanub>R7RS-small seems quite OK to me .. except maybe the weirdo exception-handling mechanisms instead of escape-continuations and delimited-continuations. :)
<taylanub>(Or maybe escape-continuations aren't "abstract" enough ?)
<taylanub>Other additions are SRFI-9, the simple, declarative library system, and bytevectors; these also seem all fine to me, as "additions to R5RS," so to say. Did I miss anything major ?
<civodul>Hello Guilers!
***mario-go` is now known as mario-goulart
<add^_>I need to figure out how to use ijp's monads to fix my haiku thingy.
<xaocon>i'm trying to build guile 2.0.7 and i'm getting a fail when i check ftw.test
<civodul>xaocon: could you try 2.0.9?
<xaocon>sure i'll give it a shot. was trying to build within my distro though
<civodul>ah, ok
<civodul>xaocon: are you running tests as root?
<xaocon>fails as both unpriv user and root
<shanecelis>hello guilers
<shanecelis>The beginnings of an Emacsy WebKit Viewer
<mark_weaver>looks nice! :)
<shanecelis>I wish I guile had some dbus bindings.
<mark_weaver>dbus bindings would definitely be a very good thing to have. IMO, it should be a high priority. wingo posted about this long ago:
<shanecelis>There is this code:
<shanecelis>But it's description is, "Incomplete and bitrotten D-Bus 0.22 bindings"
<shanecelis>LOL "Doubtful. You have some hacking to do :-)"
<mark_weaver>I wonder if any of this would be easier now with the dynamic FFI. I don't know enough about dbus to know. Maybe ask on the mailing list?
<shanecelis>Yeah, also SWIG has been updated for guile 2.0.
<davexuni`>shanecelis: that looks great!
***davexuni` is now known as davexunit
<davexunit>mark_weaver: I've found the dynamic ffi easier to use to make library wrappers.
<mark_weaver>yeah, I think that's the way to go in the future.
<mark_weaver>There are so many tricks you can easily do from scheme that would be a pain from C. Maybe that would help with the difficulties wingo faced creating a dbus binding in 2004, dunno.
<davexunit>how big is the dbus api?
*davexunit hopes for a rather small api
<mark_weaver>I've never looked at it carefully, but see for wingo's post about the difficulties he ran into.
<mark_weaver>It might be important to keep it integrated with the guile-gnome stuff. I'd recommend consulting with wingo before doing much work on it.
<davexunit>yeah, I don't think I'll be attempting this anytime soon.
<shanecelis>davexunit: thanks!
<shanecelis>I'm still super impressed with (procedure->pointer ...). It works with closures!
<davexunit>that's a good one.
<taylanub>Hrm, I was thinking I'm just wasting time with my "bytestructures" pet-project since I thought the use-cases don't come up so often, but that above blog post seems to describe a problem similar to what I'm trying to solve .. cannot tell if it's just analogous on an abstract level, or a concrete relevance.
***xaocon is now known as xaocon|away
***fanigsm-nomnom is now known as fangism