IRC channel logs
2016-05-08.log
back to list of logs
<galex-713>Also, do you think one day Guile could compile to java or C# virtual machine? <amz3>I'm trying to redirect output log done by my program using (format #t "foo") <amz3>I exec my program using `guile -L . hypermove-worker.scm > output.log' <amz3>the problem is that when I try to `tail -f output.log' in another terminal I get nothing <amz3>sorry I get something, but the output is dealayed <amz3>this is approaching a 0.1 release <amz3>it still has loose ends, but... but... but... it's scheme at it finest :P <mayuresh>a quick question. can i use guile-2.0 while learning from 'sicp'? <ecraven>ah, good times... Godly Plate of the Whale <mark_weaver>amz3: writes to a file are delayed because they are buffered by default. either use 'force-output' to explicitly flush the write buffer, or use 'setvbuf' to set your desired buffering mode explicitly. <amz3>ACTION sent a mail to guile-user <fhmgufs>Is there an formatted input method? I mean something like scanf in C. <mark_weaver>fhmgufs: not anything like scanf, but see 'string-tokenize', 'string->number', etc. <fhmgufs>What I want to do is reading a string like "Hello, I am fhmgufs" and getting "fhmgufs" as a value. <wingo>see also string-match, which is a different thing <fhmgufs>wingo: Actually that was what I found useful :). I thought that was meant. <civodul>wingo: are you looking at test-decoding-error in ports.test? <mark_weaver>wingo: from reading that, I agree that the port position should not be advanced when the error-handling-mode is raise <wingo>civodul: i haven't seen the error, i was looking at refactoring the code and noticed a thing that didn't seem right to me <civodul>this thing in ports.test checks the pointer position in the presence of errors <mark_weaver>wingo: for the 'ignore' and 'replace' error-handling-modes, iirc, unicode specifies how many bytes should be read <wingo>mark_weaver: yeah. and for iconv we have our byte-at-a-time shenanigans <wingo>but i think we do the correct thing already re: choosing how many bytes to advance <wingo>there is a comment in the source indicating the relevant unicode section <mark_weaver>wingo: okay, I wasn't sure if you were going to rewrite some of that <wingo>my question was more about whether to advance the pointer at all in the error case <wingo>i have a status mail to send soon <wingo>i managed to get a loop of 1e7 peek-char's on a utf-8 or latin1 stream to 900ms, with peek-char in scheme; vs about 380ms with peek-char in c <wingo>a somewhat frustrating difference, some possibilities to whittle away at it <mark_weaver>wingo: the text is not quite as clear as one might hope for, but given that they explictly say "bytes are ignored" for the other two cases, I think it's reasonably safe to assume that no bytes should be ignored in the 'error' case. <wingo>mark_weaver: yeah i agree. thanks for taking a look! <wingo>well, hacks for tomorrow. sleepytime!