<leoprikler>lispmacs[work]: there's also string-tokenize IIRC, plus good ol' regexp and peg ***terpri is now known as robin
<flatwhatson>ah, i see doc-snarf is in need of some love... was hoping to use this for my projects <flatwhatson>it seems to struggle with internal docstrings though, so some more attention is needed <flatwhatson>i'm thinking the way forward might be to start with adding some test cases ***sneek_ is now known as sneek
<robin>uh, i think i just killed sneek, by asking it "help later tell" <robin>i don't remember who runs sneek or i'd ping them directly, sorry about that :( <leoprikler>flatwhatson if by internal docstrings you mean the "First string in a function", then I'd rather suggest the texinfo module(s) <leoprikler>Then tell was disabled, but it seems later tell is the real bug :P <leoprikler>Or rather it appears sneek doesn't know what to do with multiple commands <leoprikler>I think 8sync might come with something, but I'm not sure <manumanumanu>I just realized guile is pretty fast starting up. On my system it even beats python3. Only perl and python2 is actually faster. <manumanumanu>RhodiumToad: I remember john asking you about some bytes->hex thingie (you have a lovely sparse array imlpementation). Did that ever materialize? Did you ever put it in a library? It would be great to have <RhodiumToad>it's not really a sparse array as a cheating way of storing 257 distinct values per byte <RhodiumToad>rather than storing byte x at index i, it stores (- x (logand i 3)) because due to the nature of the values stored, we know that x can't be less than i <RhodiumToad>and we also know that if x is 255, (logand i 3) is not 0 <RhodiumToad>so the upshot is that the value 255 never needs to be stored, so we can use it as a flag for "this value isn't valid" <civodul>when you have a URI-reference relative to some URI, what's the proper way to build the corresponding URI? <civodul>as in: URI-reference + base-URI => URI <lloda>my readline troubles are over after... reinstalling the proprietary nvidia driver. That doesn't use libedit while the mesa driver did. Haha. <sneek>Not as far as I can remember. <lloda>sneek: how far can you remember <sneek>rlb was last seen in #guile one day and 17 hours ago, saying: But I *did* like being able to write clj code for the browser (though I haven't been doing any of that for a good while now).. <dsmith>lloda: say that "help later tell" command again, and then stay quiet for a bit please. I want to see something. <sneek>lloda was in #guile 56 seconds ago, saying: sneek: help later tell. <lloda>apparently it comes through llvm <lloda>and llvm added it for convenience since a couple libraries using llvm also needed that functionality <alwalo>Sorry I am not trying to spam, but I think my messeges were invisible cause I didn't finish registration on IRC, now I finished and I send my question again (doesn't mean to spam) sorry if it was visible before <alwalo>I am using (http-request "url") without any other parameters, I do get error: web/http.scm:252:2: Bad media-type header component: "text/xml", of course I am trying to download xml file, how should I do it ? <emestee>alwalo: that doesnt sound right, are you sure the remote server is functioning correctly <dsmith-work>wingo, civodul: Might be a good idea to update the /topic to mention that your nick must be regisered to speak in the channel. <civodul>dsmith-work: oh, was it the case back on Freenode? <dsmith-work>There was a weird thing that I couldn't change my nick back unlesss I parted the channel. <alwalo>there is misspelling in last (display) call but it is because I had to retype it by hand <alwalo>I have now checked it exact same program written in Python using Requests library and everything works fine, file gets downloaded and saved on drive <dsmith-work>alwalo: I would wireshark the attempts for both and compare what is different. Possibly a missing content-type header? <dsmith-work>alwalo: Maybe somtething this is needed: (http-request "http://myurl" #:headers '((content-type application/xml))) ***sneek_ is now known as sneek
<alwalo>can you give an idea how #:headers should look like ? <alwalo>web/request.scm:133:2: In procedure validate-headers: <alwalo>Bad request: Bad value for header accept: (application/xml) <alwalo>web/http.scm:252:2: Bad media-type header component: "text/xml" <lispmacs[work]>leoprikler: string-tokenize sounds pretty close to what I want, though I'm not sure about the challenges of applying that to a live input stream in my use case <dsmith-work>alwalo: Hmm. I dunno. ((@@ (web http) validate-media-type) "text/xml") -> #t <dsmith-work>alwalo: How about this, try adding #:validate-headers? #f to your http-request <dsmith-work>Sorry, looks like you can't pass that to http-request <alwalo>(read-request-body (build-request "url" #:headers .....)) should this work like this ? I will try <lampilelo>lispmacs[work]: for live string input streams maybe 'expect' would be of interest to you <lispmacs[work]>manumanumanu: just that I'm reading from a port in real time connected to a device that delivers whitespace separated tokens that need to be processed one a time in real time <lispmacs[work]>I'm pulling output from a FORTH interpreter running on a microcontroller <manumanumanu>sure thing. srfi-171 does that, but you will need to write your own reducer for it that writes to a port ***lispmacs[work] is now known as forthmacs[work]
<forthmacs[work]>It is becoming harder and harder to find a telephone booth these days ***forthmacs[work] is now known as lispmacs`
***lispmacs` is now known as lispmacs[work]
<manumanumanu>lispmacs[work]: I am not sure how to do it best, though. Unless you want to do some kind of processing, transducers seem like a waste of processing power. <lispmacs[work]>manumanumanu: this Expect module might be what I was looking for <manumanumanu>but that unhygienic capture of variables is pretty unsexy though. ***Server sets mode: +ntz