<amz3>I'm having a look at opencog, I was wondering what you know/think about PLN <amz3>the acronym means probabilistic logic network <amz3>I did not read much about the subject, anyway it seems to be somekind of markov chain framework <amz3>it relates to markov chain somehow <amz3>since opencog database is basicaly a graph, i think that they somehow how have something to infer knowledge by doing something like guile-log does but use probablity <ArneBab>paroneayea: *blush* — that blog entry is wow! <taylanub>"Now what would really be hylarious would be to port [Wisp] to Hy..." definitely :D <lloda>what does it mean when ,profile gives me a (%, cumulative, self) line like [1.16 194.01 (!!) 0.15 srfi/srfi-1.scm:590:5:map1] even though the total time is ~13 seconds? <ArneBab>besides: I already did changes to catch the locale errors and always use bash in bootstrapping, and I now also include wisp.scm in the tarball. I just have to create a release for that. <ArneBab>I fear though, that I have some unpushed work for that lying on the other computer. <ArneBab>…well, won’t kill me. If I have, I can still merge it later and release it as a further version. <paroneayea>too bad there's that name collision, I wonder if people will think of the other Wisp <paroneayea>ArneBab: cool! I'll look at updating the Guix package submission later. <ArneBab>but my adding of wisp.scm did not work… <ArneBab>paroneayea: I now released 0.9.0: it no longe depends on Python for building <ArneBab>civodul: I wanted to answer, but could find no words which would have expressed the jadedness I felt ***karswell` is now known as karswell
<zacts>but I still prefer parenthetical syntax <taylanub>same here, if just for paredit. Wisp looks much nicer than SRFI-110 I have to say. <sbp>wisp is the future, people <ArneBab>taylanub: keep in mind that wisp is *almost* SRFI-110 — the differences are much smaller than the similarities. <ArneBab>(mostly the differences are leaving out features) <ArneBab>the : syntax grew out of the \\\\ syntax in SRFI-110 <ArneBab>the _ as optional prefix for lines to survive HTML grew out of the ! in SRFI-110 <ArneBab>so the only really new thing is the . as prefix for lines without parens <ArneBab>(which is at the same time what I like best and what I expect will prove most confusing) <ArneBab>the rest is “let’s to some letter statistics — the most common unused symbols are : and _” <davexunit>sbp: I disagree, but Wisp is certainly interesting. <sbp>civodul is the future, people <ArneBab>it will have to prove with usage that there are no important patterns which hit the nasty dark corners of wisp :) <ArneBab>(I defend SRFI-110, because it’s unclear whether my choices for wisp as opposed as the choices of SRFI-110 will actually prove to be better) <ArneBab>(wisp feels as a system should feel, and thanks to being based on guile — and only ~1000 LOC — it’s likely to survive long enough that it can get enough testing and hopefully infrastructure, but I only have one and a half decades of programming experience, and systems from far better people fell flat when they encountered sexps) <sbp>civodul: sorry, was referring to context from just before you joined <sbp>in summary, just an admiration of your work <ArneBab>(but then: if it should fall flat, all needed to salvage the work is `for i in *.w; do guile wisp.scm "$i" > "$(basename "$i" .w).scm" ) <ArneBab>and I fully agree with sbp: civodul, you’re doing awesome work! <davexunit>paroneayea: I started checking out squee last night, but I only had ~10 minutes to look at it. <davexunit>the first thing I thouht of doing was removing the paramstring connect procedure from the public interface and wite a procedure that accepts either an alist of config values or a uri object. <civodul>and it needs an EDSL for SQL queries :-) <davexunit>civodul: yes, a notoriously difficult problem. <davexunit>I actually just fixed a sql injection vuln in a rails application :x <davexunit>and from a language design standpoint, there's lots of decisions to make: s-exp based, or something with record types? how does one compose simple sql expressions to build more complicated ones? <davexunit>do we attempt to use relational algebra as some have tried elsewhere or go a different route? <daviid>paroneayea: davexunit I'm interested in participating in squee, I also have immediate suggestions about the structure of the project [source tree] and I'm hapy to immediately autotool chained it. Last, I really think it should be on savannah, I'm happy to do the admin prep/quizz if you agree. I actually think we should get it ready for an official GNU Project, I can do that too <paroneayea>davexunit: the paramstring thing is built-in to postgres though <davexunit>paroneayea: yeah, that would be the low-level procedure we'd call <paroneayea>davexunit: how would an alist work though? I don't remember that you can reference things by keys <davexunit>paroneayea: the param string has a format like "key1=value1 key2=value2 ..." <daviid>i just took the oportunity to ping both of you wrt my interest ... <paroneayea>daviid: I'm fine giving you commit access; you and I haven't worked on anything together yet, but you seem to do good work in here, and squee is pretty early stages... I usually don't give out commit access this early on to somoene I don't know well but I think I'm okay with it here, and I've heard good things about your guile-gnome stuff <paroneayea>daviid: also re: being on savannah, sure I suppose, I don't have that strong of opinions at the moment if you want to take the work to do it <paroneayea>daviid: autoconf'ing it up will be ggreatly appreciated, not my favorite thing to do! <paroneayea>daviid: as for technical direction, I'd like to coordinate before making major decisions. In particular, at the moment guile-squee is verrrry minimalist and low-level, and I'd be happy to expand beyond that point, but looking forward to what that is, I think needs some talking about <paroneayea>daviid: I'd like to keep the low level of it functional, though I'd be happy to have some optional goops-oriented wrappers higher up, as long as they're really optional, kind of how sqlalchemy has sqlalchemy core and sqlalchemy orm <daviid>paroneayea: thanks! I'm sure it will be a very fruitfull colaboration, and agreed with all you said above. <daviid>tell me when you have set write priv, I will create a deval branch and autool first things first <daviid>then I will ask to be an official GNU Project, if ok perfect, if not a nongnu savannah ... <paroneayea>but I think that's kind of future stuff, if ever :) <daviid>paroneayea: I looked at these following your posts here. <paroneayea>I'd like to do something like what racket's database stuff does <paroneayea>daviid: so as for the GNU project thing, I have no objections, but do you think it's promising enough to shoot for that? <paroneayea>I haven't put enough time into it to be totally confident it'll succeed yet <daviid>paroneayea: i have to concentrate on something else now, but I agree and will halp you and davexunit so we get a good sxql like approach, I really like it as well <paroneayea>yeah I think you davexunit and I might be a good team to getting that to happen, as I'm sure even if davexunit has tons else on his place he'll take verrrrry curious interest in that kinda thing :) <daviid>paroneayea: I don't have a noteabug account <taylanub>ArneBab: interesting. still it makes a huge difference in impression for me. <*> and ! and \\\\ and $ look very ugly to me, to the point I feel like going "just say no" against SRFI-110 :-) the visual cleanliness in Wisp is very attractive in contrast. <paroneayea>daviid: oh, well if I'm giving you commit access, until it's on savannah, so maybe you should set one up? <daviid>do I have to have one to be committer ? <paroneayea>it's all free software and run by free software people though <daviid>ok, will create an account and ping you, give me a sec <davexunit>hoping the next Emacs maintainer has good feelings about guile-emacs <paroneayea>Stefan has been sometimes a bit rocky but overall has served emacs well <taylanub>ArneBab: by the way how feasible is automatic transformation between regular sexpr and Wisp code, both with idiomatic indentation/formatting? the thought of having to work with other people's Wisp code worries me a bit because I'm used to editing with Paredit and reading parenthesized code. <paroneayea>daviid: added as a collaborator... give it a go! <davexunit>paroneayea: my notabug username is the usual: davexunit <paroneayea>davexunit: daviid: where should we do collaboration for now on issues, etc? <davexunit>paroneayea: guile-user would be appropriate for discussions <davexunit>especially since knowledgable folks could chime in <davexunit>that aren't directly involved in development <paroneayea>maybe I should send an email about squee to that list <daviid>paroneayea: davexunit once i will have asked for a savannah either officla gnu or not, will have a mailing list <daviid>but for now, guile-user is our best choice imo <daviid>how do I clone as a commiter on noteabug, do you know? <ArneBab>taylanub: for me it’s the same with visual cleanliness — visual impression does matter <ArneBab>taylanub: for SRFI-110 there’s a script named sweeten, but I don’t have one for wisp. It would be great to have, but non-trivial to get *good* <paroneayea>daviid: see that little cloud dropdown box on the guile-squee page? <ArneBab>taylanub: the conversion wisp->regular scheme is clear, but requires string parsing to keep the comments. <daviid>ok did you add my ssh key already ? <taylanub>ArneBab: it would be very very neat if I could just open Wisp files and have them appear as regular Scheme code transparently. (I have an Emacs mode doing that for JSON, representing it as sexprs, for instance.) <ArneBab>functional conversion is easy, but complete conversion is harder <taylanub>ah, I see the comments might be problematic <paroneayea>daviid: what name do you go by btw, aside from daviid? :) <davexunit>taylanub: whoa what emacs mode does that for JSON? <daviid>paroneayea: you mean my real name? David Pirotte <taylanub>oh, the README might be a bit outdated ... IIRC this currently requires the modded json.el there. I usually work on this in brief bursts in the middle of other tasks; should sit down and clean up the json.el compatibility issue. <ArneBab>taylanub: what it might be able to do is adding an overlay which simply adds parens and replaces : by an opening paren and a closing paren at the end of the line <ArneBab>but for using paredit it would need a wisp-equivalent of that <ArneBab>all in all that requires a wisp-parser in emacs lisp <ArneBab>essentially it’s just a theoretically simple transformation, but there are lots of nasty corner cases <paroneayea>ArneBab: might be fun to have a wisp-parser that can even allow one to code emacs lisp in wisp :) <ArneBab>(I mostly stumbled over parsing of strings <paroneayea>I do like that wisp's syntax is pretty "generic" for anywhere sexps can live <ArneBab>wisp.scm should be able to do it for most elisp code <ArneBab>it’s just a representation of arbitrary tree structures <ArneBab>(plus lots of hacks for special cases like #'() and such ☺) <ArneBab>the code actually doing the transformation into a tree isn’t that big <ArneBab>actually all the code isn’t that much :) <davexunit>paroneayea: ah this indeed went unnoticed for me. <ArneBab>(I just realized that in my mind the special cases are much bigger than when I count the lines ☺) <davexunit>we're going to put a lot of miles on this feature, I think. <paroneayea>we should rebrand it as guile-mapreduce and get all the hipsters to join in <daviid>paroneayea: all set, I just created and pushed a devel branch [did not add anythig in it yet, just wanted to make things work, all fine] <daviid>paroneayea: davexunit thanks! it will be very interesting to work together <davexunit>we're all busy, so if we can all slowly chip away at this together, we should get farther than we could have alone. <daviid>davexunit: i agree, all very busy, and we will do more 3 of us then alone... <paroneayea>I even said I prefer parens in the post personally, but that I think it's a useful syntax to help people become accustomed to lisp <daviid>how to define a module and export things using RS7RS, any quick tip/example? <taylanub>daviid: I assume you mean R7RS. Guile doesn't have full support for R7RS yet, but you can just use native Guile modules. <daviid>i just don't know how to define nd use a module <taylanub>oh, I see you asked in #scheme too, so let's move there <daviid>(define-library ij)\\n(export ij-open)\\n(define (ij-open filename)\\n(ij.IJ:openImage filename)) <daviid>ij.scm:35:1: invalid list in library name <daviid>yeah, sorry for the noise here folks! <amz3>the path I've taken is very difficult <amz3>too much features at once