<davexunit>ACTION tries to see if there's some way to encode additional information in a texinfo file that will allow me to perform syntax highlighting on the text in an @example block. <ArneBab>davexunit: maybe javascript-based highlighting? <davexunit>I'm definitely going to avoid javascript highlighting, though. <cmhobbs>can anyone point me to documentation for how i might open a tcp port with guile and send/receive strings from it? <wleslie>if you want to send and receive strings over TCP, the first thing you'll need to do is decide upon a message framing format <wleslie>because TCP is not a message protocol, it's a stream protocol <cmhobbs>i'm trying to communicate with gnu go in gtp mode <cmhobbs>i've done this with ruby with the TCPSocket class and the puts method <cmhobbs>do i need to handle lower level stuff by hand in guile? <wleslie>cool, so you've already got a format for the messages <cmhobbs>7.2.11.4 Network Sockets and Communication looks promising from the manual <wleslie>you'll need to convert between bytevectors and strings, too <cmhobbs>i'm not sure i need to. i'd just like to connect to a port <wleslie>yeah, you can just use the port interface <cmhobbs>i'll dredge that up in the guile manual <cmhobbs>im digging through the socket stuff right now <mark_weaver>you can write and read text strings directly to/from the socket port, so there's probably no need to use bytevectors. <cmhobbs>mark_weaver, where should i dig in the documentation for that sort of thing? got any useful function names i can poke at? <cmhobbs>i'm still getting started with guile so i don't have the docs <cmhobbs>i'm looking into ports and file descriptors now <mark_weaver>the link that wleslie gave above covers the socket side of things <mark_weaver>after that, it's the same as I/O to any other port type. <cmhobbs>thanks! i've been struggling to find time to get guile under my belt, so it's coming in fits and starts :D <cmhobbs>i'll have some downtime tomorrow, so i'm trying to prep some docs <roelj>Is there anything special I need to do to get (procedure-source) to work? <amz3>funny it doesn't seem to work <amz3>ACTION digging the source <amz3>procedure-sources is really program-source <amz3>is defined guile/module/system/xref.scm <roelj>amz3: Ehm, can I in some way treat a procedure definition as a list? <roelj>(I would like to treat it as a syntax tree actually) <amz3>roelj: see define-syntax & syntax-case <taylan>roelj: what exactly is it that you're trying to do? (from a higher-level perspective) <roelj>taylan: I have a function with decision-making criteria (which boils down to a bunch or predicates). When the program is running, I would like the program to be able to alter the predicates (which I'm limiting to adding and/or removing predicates). <taylan>roelj: maybe you could keep the predicates in a data structure, like a list, vector, or table? <roelj>taylan: Yes, that would be awesome. But how can I store unbound variables in them? So when I have: (< a +inf.0), then how can I represent the a (which will be bound later) in the list? <roelj>Should I use a macro for that, like amz3 said? <amz3>taylan knows better. I think you should use procedure to bind the unbound variables at runtime ((lambda (a) (< a +inf.0)) 42) <taylan>roelj: hmm, I don't understand how that relates to keeping the predicates (not variables) in a data structure. can you elaborate? normally, you can just put e.g. (lambda (a) (< a +inf.0)) into the list, and later just call it on some value. (probably every predicate in the list should take the same number of arguments...) <taylan>very small example system: (import (srfi :1)) (define predicates '()) (define (passes-all? x) (every (lambda (pred) (pred x)) predicates)) later: (push! predicates (lambda (x) (< x +inf.0))) (push! predicates (lambda (x) ...)) (passes-all? 5) => whatever <taylan>(push! isn't defined in the standard environment. a proper data structure might be better anyway though.) <amz3>roelj: are you building a decision tree? <amz3>ACTION wonders how he wrote so much code <roelj>amz3: Yes I suppose I'm writing a decision tree. And I want to add new conditions to the decision tree while running the program. <roelj>Let me try to write a small example code <davexunit>alright, I implemented a way to highlight code blocks in texinfo documents for my blog. <amz3>you use texinfo for the source of your blog? <davexunit>I was experimenting to see how well it worked. <amz3>what's the advantage of texinfo over skribe? <davexunit>the advantage is that it is nicer to write in most cases <davexunit>emacs already has a really nice mode for working with it <davexunit>the disadvantage is that I have no way to insert code when the need arises. <amz3>I think pollen parser is easier to write that markdown because it has an escape sequence <davexunit>the issue I have with Pollen is that you write plain HTML or whatever else <davexunit>Skribe is better in this sense, but in practice I'm finding Skribe hard to work with. <amz3>maybe I'll end up fixing combinators to work like I want, and add pollen syntax to it <davexunit>perhaps mostly due to missing the necessary elisp to make it pleasant. <amz3>I don't have that requirements <amz3>and did not write much skribe either, i did not seem bad <rekado>writing a blog in texinfo format is a neat idea. <davexunit>rekado: it totally works! guile already has built-in texinfo stuff which works great. <davexunit>I just needed some extra pre-post-order SXML transformation to find 'pre' tags and highlight the code within them. <davexunit>the big downside being that you can't insert scheme code within them. <rekado>guile-emacs segfaults for me when I load my full configuration file. <taylan>my .emacs also has problems. I should learn how to use GNU C debugging tools properly and get to looking into things... <davexunit>you can play the ol' configuration bisect game to find the offensive code <rekado>I've been playing this already and it seems that it's nothing in particular that causes this. Just *anything* after the point where I'm loading up magit seems to crash it. <rekado>I had to remove a couple of things from the config already, but those were scheme exceptions, no segfaults. <rekado>or maybe the segfault has nothing to do with my config but with the attempt to restore a desktop. <rekado>yup, there is no segfault when I comment the expression (desktop-save-mode t) <sneek>Welcome back daviid, you have 1 message. <sneek>daviid, ArneBab_ says: arne_bab -ät- web -dot- de ***dje is now known as xdje
<amz3>daviid: how clutter is going? do you need help to test? <daviid>amz3: clutter is fne, but if you want to help, you could can guile-lib and try to help me solve the make check 1 test failure, that would be awesome <amz3>daviid: the error is in api.scm ? <daviid>oh, it pass! i'm working on a devel branch and made modification to the autool config. the failure was in texinfo.serialize.scm <daviid>davexunit: for info, ssax ssax-simple are guile-lib sxml modules <daviid>amz3: I can not reproduce that here, and the devel branch [which I did not push my modifs yet] pass amke check now <daviid>you copied part of the trace, did it make fine fails make check or fails make ? <amz3>make check fails with the backtrace I pasted <daviid>ok, then let's wait i finalize the modifs on the devel branch, then you try again [not today though] <karhunguixi>i'm working my way through a tutorial and am puzzled at why i can do: "(map 1+ (iota 5))", but not "(map 2+ (iota 5))" <mark_weaver>karhunguixi: '1+' is just a name like 'map' or 'iota'. <mark_weaver>and it so happens that '1+' is bound to a convenience procedure that adds one to its argument, but no such binding for '2+' is provided. <karhunguixi>amz3, that (equals?.. code fails for me, what is it supposed to illuminate? <amz3>it means that 1 + 2 == 3 <amz3>equal? is the procedure to test equality by value <amz3>in this case, eq? can replace equal? <taylan>eq? shouldn't be used for numbers. (eqv? is fine.) <taylan>eq? is an efficiency hack over eqv?. for anyone interested in the precise equivalence semantics of Scheme, it's best to try to understand eqv? first. <taylan>the terminological distinction between "equivalent" and "equal" is not that clear I think. I'm guessing it's due to some historic reasons, but not sure. <taylan>eqv? tries to approximate the concept of "operational equivalence" though, maybe it's from there <karhunguixi>ok, it's of secondary importance to me. I like to know what things spell out, even if it's not correct (anymore) <mark_weaver>karhunguixi: for numeric comparisons, it's usually more appropriate to use '=' <mark_weaver>and (= 0.0 -0.0) is true, but (eqv? 0.0 -0.0) is false <mark_weaver>eqv? tests for "operational equivalence", whereas '=' tests for numerical equality. they are subtly different concepts. <mark_weaver>you need operational equivalence to implement memoization, but in most other cases '=' is what you want. <ArneBab_>karhunguixi: (1+ 2) is equivalent to (+ 1 2). You can define it yourself with (define (1+ x) (+ 1 x)) <paroneayea>I like that The Art of the Propagator has a small footnote at one point, simply saying <karhunguixi>ArneBab_, what would you call this? maybe procedure overwriting? <rekado>karhunguixi: "procedure definition". With 'define' you are binding a procedure to a name in the top-level environment. <karhunguixi>what about when there's already one with the name you specify <rekado>you can try this in an interactive session. <rekado>it's probably not a good idea to do this at the top-level, though. <karhunguixi>yeah, i've tried this. Was just wondering in general if there were any special procedures that's not possible to redefine <rekado>you can even shadow things like "if". <rekado>I don't know if anything is really sacred :) <rekado>this fails: (define (define x) x) <rekado>but only because "define" in the subexpression is expanded first (and then triggers a syntax error). <rekado>but "define" is not a procedure, it's a special form <rekado>"shadowing" is "covering" a previous definition with a new definition. <rekado>it is used when a name is bound in an outer scope as well as the current scope. <rekado>the binding at the current scope is said to "shadow" the other binding(s) (in this scope). <rekado>(let ((a 10)) (let ((a 20)) (display a))) <rekado>the inner binding of "a" shadows the outer binding; "20" is printed rather than 10.