<sneek>Welcome back daviid, you have 3 messages. <sneek>daviid, str1ngs says: thank you I would start with this example. <dsmith>Does anyone remember some Guile code that would make a dot dependency graph of which modules use which modules? <dsmith>I can't seem to find it in the guile sources. Or the scwm sources. And I *know* I was using something like that at one time.. <dsmith>Was tracking down circular dependencies in scwm. <daviid>dsmith: I can't help, though I have a (very) vague memory of this being mentioned here, a (few? :)) decade(s) ago, not sure <dsmith>grep and sed and echo to the rescue. <dsmith> echo 'digraph g {' ;grep ':use-module (app scwm' *.scm | sed -n 's/.scm:.*#:use-module (app scwm//p' | tr -d \) | sed 's/ / -> /'; echo '}' ) > ~/deps.dot <daviid>dsmith: if scwm is in guix, guix has one, afaict, i still don't use guix :) <daviid>oh, you mean strict guile scheme modules, forget about mu suggestion then ... <daviid>str1ngs: could you please repost, the debia pate expired, sorry ***jonsger1 is now known as jonsger
<str1ngs>my closure might be wrong. it replicates the C signature. though passing NULL or #f would be enough for now. <dsmith>daviid: Yeah, just within scwm. It's old. From around Guile 1.3.5 I think. <apteryx>what's the most idiomatic way to solve a problem where I have two same length lists, and one of the lists contains a boolean that "enables" or "disables" the values of the second list? The goal is to recreate a list with only the "enabled" values. <dsmith>I'd look at filter. Can filter take multiple args like map? <apteryx>dsmith: yikes. This used to happen to me a lot because of my use of ratpoison (lack of any notification system). <apteryx>now I have an mcron job that beeps the battery before it does ;-) <apteryx>RhodiumToad: fold-right. Will read on it. <dsmith>scheme@(guile-user)> (use-modules (srfi srfi-1)) <dsmith>scheme@(guile-user)> (filter-map (lambda (p v) (if p v #f)) '(#t #f #t #f) '(a b c d)) <dsmith>Only tricky thing, is if your "value list" has any #f values, they will be removed.. <RhodiumToad>that'll be a problem if any element of the list is itself #f, though? <dsmith>In Haskell, I'd try some combo of zip filter and map. But I'm just learning. <daviid>str1ngs: the problem is that g-golf still misses an implementation for callback GIArgument types <daviid>str1ngs: for info, when the type-tag of an argumet is interface, currently g-golf handles enum, flags, struct and object, so I need to work on this to also handle callback - for info as well, you may look at (g-golf hl-api function), prepare-gi-args-in (line 562 ...), then the match type-desc starts line 580, and fails since there is no 'callback type-desc code yet :) <chrislck>(fold-right (lambda (a b c) (if a (cons b c) c)) '() p v) <daviid>str1ngs: i run your example, manually scroll down and how funny: on the bottom right of the page, "Today's random package…" Guile-gome :):) <dsmith>So no one does corba with gnome anymore, right? <daviid>str1ngs: it might be interesting to display the 'full url' 'somewhere' when we hover over a link (?) <daviid>dsmith I don't know, but isn't it that corba has been deprecated, not sure <daviid>str1ngs: because of this missing callback type, even if yu pass #f or %null-pointer, it will raie an exception ***jonsger1 is now known as jonsger
<str1ngs>davexunit: thanks for the feedback. for now I'll use a typelib function when I need to call javascript. <str1ngs>daviid: hovering over a link my actually be more of a challenge then you might thing. unless webkit has some primitive for this. Which as far as I know it does not. you would have to use DOM introspection. which requires either javascript injection or IPC via a web extension. javascript would need the missing GAsyncReadyCallback at the very least. for nomad I'm hoping to add some kinda DOM introspection that uses scheme, but I think <str1ngs>daviid: if you think of anything in regards to DOM introspection. let me know it's not going to be an easy problem to solve I think. ***lavaflow_ is now known as lavaflow
<lloda>scm_array_handle_bit_elements_offset is undocumented (?) <lloda>oh it's mentioned, just no heading and not in the index :-/ ***dddddd_ is now known as dddddd
***matijja``` is now known as matijja
<bandali>dsmith-work, are you the author/maintainer of sneek? <bandali>dsmith-work, cool! would you consider changing the "tell" functionality to default to "later", and only relay the message immediately with "now"? <bandali>it's a somewhat regular occurrence over at #guix :-) <dsmith-work>bandali: Yes, I'd like to fix that. Its just very lightly warmed over port of the old sarahbot that used to inhabit #scheme <bandali>dsmith-work, ah, thanks. i'd be happy to send a patch if it's free software. if not, i'll leave you to it :-) <dsmith-work>Which was running on some JVM based Scheme. (can't remember at the moment) <dsmith-work>bandali: I'd like to release it, but I have a lot of work to do on it first. And i've just been busy with other stuff. <bandali>dsmith-work, awesome! and gotcha, no worries. would appreciate a ping if/when you get around to doing it <daviid>str1ngs: I want to fix the missing gi-argument and g-value array types first - and while working on it, I can look at and fix 'implemented feature' bugs - then I will look at the callback gi-argument type <daviid>str1ngs: wrt webkit2 'hover feedback', I know next to nothing about webkit2 and even less about DOM ... but I'd look at and snarf some of the epiphany-browser code, which does exactly that ... my 2c <str1ngs>daviid: regards to g-value array etc . thanks for looking at that. meantime it's not a huge rush. for types g-golf can't handle right now I just call via my typelib. <str1ngs>daviid: epiphany uses a web extension . which means a client/server model using DBUS <daviid>str1ngs: you should be able to just introspect a WebKitDOMDocument <str1ngs>daviid: yep, but that's only available from the webkit process when it loads an extension. which is the PITA part ***ng0_ is now known as ng0
<civodul>i see "anon" entries in (statprof) profiles and i wonder in which case we should expect Guile to lack source location info ***janneke_ is now known as janneke
<lfam>Agreed, great talk from wingo! <civodul>it's good to hear a reflection on the user base, handling changes, etc. <civodul>i tend to be conservative, but i think Guile 2.0 upwards has done a great job at handling changes <civodul>the 1.8 to 2.0 was a bit more bumpy :-), but that was probably unavoidable <dsmith-work>civodul: Yes. Getting the bot running on 3.0 was not near as traumatic some of the previous efforts. <civodul>changing GCs would have the potential to be traumatic, as wingo mentions