sword865 has quit [Quit: Connection closed for inactivity]
englishm has joined #racket
englishm has quit [Excess Flood]
notzmv has quit [Read error: Connection reset by peer]
pie_ has quit [Ping timeout: 252 seconds]
Lowl3v3l has joined #racket
liberiga has joined #racket
endobson has joined #racket
orivej has joined #racket
manualcrank has joined #racket
vraid has quit [Quit: Leaving]
ng0 has quit [Quit: Alexa, when is the end of world?]
ng0 has joined #racket
Sgeo has quit [Read error: Connection reset by peer]
Sgeo has joined #racket
badkins_ has joined #racket
badkins has quit [Ping timeout: 244 seconds]
efm has quit [Read error: Connection reset by peer]
efm has joined #racket
FreeFull has joined #racket
<casaca>
Hi. compiling a racket snapshot (with --enable-csdefault) fails with:
<casaca>
Exception in compile-file: failed trying `make-arity-wrapper-procedure`; probably you need a newer Chez Scheme
<casaca>
even tho the chez repo is freshly cloned
dancek has quit [Ping timeout: 276 seconds]
liberiga has quit [Ping timeout: 260 seconds]
pie_ has joined #racket
<remexre>
dTal: huh, so this would only work for compiling to Racket, right? I'm writing a compiler that spits out a specific IR that Racket doesn't itself support, so at least the reader -> expander flow wouldn't be as useful
<remexre>
that certainly is neat, though
dancek has joined #racket
badkins_ has quit [Ping timeout: 245 seconds]
davidl has quit [Ping timeout: 276 seconds]
acarrico has joined #racket
rain2 has joined #racket
<dTal>
remexre: that's right - the advantage if you do do it that way is your new language is interoperable with all the other racket #langs
<dTal>
mind you, once you've "compiled to racket", there's no reason you couldn't then tell racket how to generate your specific IR
rain1 has quit [Ping timeout: 272 seconds]
mSSM has quit [Quit: WeeChat 2.5]
<remexre>
The conversion would need to be able to translate all of Racket though, right?
<remexre>
dTal: ^^
<dTal>
no, what I meant was, you have this chain of syntax -> s-expressions -> well-formed racket, with lots of infrastructure to help you - there's no reason why the "well-formed racket" at the very end couldn't be generating your IR, instead of just calling semantically equivalent racket functions
<dTal>
i.e. you can write a compiler in racket and still take advantage of its #lang features, if I'm not mistaken
<dTal>
so no, you wouldn't have to target all of Racket to your IR, though that would be the nicest way
sauvin has quit [Ping timeout: 244 seconds]
<dTal>
(disclaimer: I don't know what I'm talking about)
efm has quit [Quit: Konversation terminated!]
davidl has joined #racket
<remexre>
hm
casaca has quit [Ping timeout: 246 seconds]
alanz has quit [Ping timeout: 258 seconds]
casaca has joined #racket
zipper has joined #racket
endobson has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
liberiga has joined #racket
casaca has quit [Ping timeout: 248 seconds]
<zipper>
Does storing ASCII text data save memory when worked on as bytestring instead of as string or does it just help cut away an encoding and decoding step?
casaca has joined #racket
casaca has quit [Ping timeout: 248 seconds]
ubLIX has joined #racket
casaca has joined #racket
casaca has quit [Ping timeout: 268 seconds]
casaca has joined #racket
vraid has joined #racket
ubLIX has quit [Quit: ubLIX]
<greghendershott>
zipper: I think bytes vs. string is mainly about whether the bytes are actually valid ASCII/Unicode; if they are I suggest using a Racket string.
<greghendershott>
I think Racket strings are stored internally at UTF32, so to answer your question about space, yes if you know it's ASCII then bytes will be less than string.
<zipper>
and save the encoding and decoding
<zipper>
The bytes are valid ascii and unicde
<zipper>
Just valid ascii actually
<greghendershott>
Sure if you will just hold the ASCII text as opaque bags of bits that you won't process, you could just use bytes.
<zipper>
greghendershott: Well I will perform string comparison and other such stuff
<zipper>
but it's strictly all ASCII
<greghendershott>
It's up to you. My preference, I'd start by using Racket strings, until/unless I knew I did have a space and/or speed problem (and knew I would not have any internationalization problem).
efm has joined #racket
<greghendershott>
Anyway sorry to editorialize. Your original question: yes, bytes will avoid codec and use less memory.
efm has quit [Ping timeout: 245 seconds]
vraid has quit [Ping timeout: 252 seconds]
zipper has quit [Ping timeout: 248 seconds]
Lowl3v3l has quit [Remote host closed the connection]