IRC channel logs
2015-12-07.log
back to list of logs
<paroneayea>ACTION runs into: ERROR: Throw to key `vm-error' with args `(vm-run "Unrewindable partial continuation" (#<vm-continuation 5a50460>))'. <paroneayea>I'm guessing it was because I used either call-with-values or because I used dynamic-wind, but which one caused it I'm not sure. <paroneayea>can't use dynamic-wind and delimited continuations, it seems! <marusich>I'm having trouble understanding why this evaluates to #f (in Guile): (eq? '(a b c) (list 'a 'b 'c)). My REPL prints out identical results when I evaluate '(a b c) and (list 'a 'b 'c). In both cases, it prints: (a b c). I don't understand this behavior; what am I missing? <davexunit>marusich: eq? tests if 2 objects are actually the same reference <davexunit>in other words, they are equal?, but not eq? <davexunit>paroneayea: that's where sly got it's coroutine implementation from :) <daviid>I confirm that guile-gnome make check pass when built using gcc 5.2, -fno-strict-aliasing is not required <daviid>I'll do the git house keeping and release tomorrow or the next day though, I'm too tired now <daviid>and guile-clutter grip-clutter work like a charm when the all chain was compiled against gcc 5.2 as well, great! <lloda>I use (define (format! s . args) (apply format #t (string-append s "\\n~!") args)) <civodul>lloda: i usually make such procedures a macro so that -Wformat works <lloda>ah, so that's why it gives me bogus warnings <civodul>about the non-literal string, i guess <lloda>I get warnings about argument count, actually <amz3>hmm warning about the format I guess <taylan>amz3: about possible misuse of the 'format' procedure <amz3>taylan: do you know which procedure like read/write that doesn't block? <taylan>doesn't (info "(guile) Input and Output") have something? <amz3>char-ready? seems to be the procedure I'm looking for... I'll try again <lloda>make[2]: Entering directory `/home/danielo/sys/guile-git/obj4/bootstrap' <lloda> BOOTSTRAP GUILEC ice-9/eval.go <lloda>Warning: Unwind-only `stack-overflow' exception; skipping pre-unwind handler. <lloda>building master ---it's strange b/c it worked with gcc 5.2, but I tried compiling with gcc 5.3 and I got this, then I switched back to 5.2 and I got the same error again :-/ <amz3>I can't find bv-take, bv-drop, or bv-append in the procedure-index, what I am doing wrong? <davexunit>one doesn't typically work with bytevectors in that way <davexunit>they are trivial to implement using bytevector-copy!, though <amz3>yes... I can implement them but what's the correct way to work with bv without them then? <davexunit>I work with bytevectors a lot, and I've never wanted those. <davexunit>reading binary data from a port, matrices, etc. <amz3>because I waiting for message with delimiter, if there is no delimiter i append the bytevector, otherwise I split the bytevector <amz3>I'm using the fast path u8-list->bytevector and bytevector->u8-list :o) <amz3>I should use bv-copy! though <amz3>the problem is that I need to call utf8->string at some point <davexunit>if so, it doesn't seem like you want to be using bytevectors <amz3>I'm writing a non-blocking read and write <amz3>I know it's a toy, but it will be helpful for me <amz3>davexunit: I'm trying something else, instead of read-char, I use read and catch the error <amz3>hopefully the port remains at the beginning of the failling string <amz3>the read/ procedure is so ugly <paroneayea>looks like I can't use guile's dynamic-wind with 8sync because it will result in an unrewindable continuation <davexunit>I'd be interested in what wingo or mark_weaver or civodul have to say on the subject <paroneayea>davexunit: yeah I wanted to use dynamic-wind to close ports even in case of error when leaving the web handler <paroneayea>right now I have code wrapped by a catch-all that runs at the end of the function, or if caught, runs inside the catch and then re-raises the exception <mark_weaver>paroneayea: we could probably reimplement dynamic-wind in pure scheme, but forget about using slib for this. we'll have to make it for guile specifically <mark_weaver>e.g. it will have to integrate with the C implementation of dynamic-wind <mark_weaver>alas, I don't have time to explore this further right now <mark_weaver>more generally, we would like to reduce the amount of C code we use in Guile, for many reasons. the most important parts to rewrite in scheme are the procedures that call back into Scheme, so that their activation frames end up in the middle of the stack. <daviid>for info, Guile-Gnome 2.16.4 released <paroneayea>is it really possible at all to dump it into an sxml document as-is without being escaped?