IRC channel logs

2026-04-03.log

back to list of logs

<dsmith>old, The BoF at 9:00 EDT is AM ?
<old>yes
<dsmith>Thanks
<old>I should have put both 12/24 formats!
<dsmith>The 13 UTC gave me a good hint you were talking 24, but I wanted to confirm
<kimothy>Hi! Anyone in the room familiar with GNU Artanis? I´m struggeling to use sqlite as a db, so hoping somone has a tip :)
<lechner>Hi, for exception handlers, why isn't #:unwind? #t by default, please?
<identity>lechner: the standards specify the behaviour you get with #:unwind? #f; besides, you do not always want to unwind, or perhaps do so manually
<jcowan>identity: which standards?
<jcowan>lechner: the basic answer is that resumption semantics is inherently more powerful
<identity>i think at least R⁷RS; you need to call/cc out of the handler to unwind
<jcowan>R6RS as well, actually
<rlb>found the (a) segfault! I would really like some flavor of (particularly, at least, dev/test time) bounds checking for C-ish (surprise). Said no one else ever...
<identity>you can sprinkle around some asserts, though that does not save you from the syntax. nothing is going to check your a[b]
<rlb>As you might imagine, all the utf-8 changes have excellent potential for trouble there -- not just via bounds, but via often two flavors of bounds, char-count vs byte-count for any given buffer...
<rlb>anyway, happ(y|ier|ish) now
<rlb>I know there's some work wrt "counted by" support going on these days, but haven't followed it closely, and don't know how compiler-specific that is, or whether it'd even be relevant to this case.
<jcowan>identity: I misread what you said: I thought you were saying that "#:unwind #f" is standardized, but you meant that the *semantics of* "#:unwind? #f" are standardized.
<n|Phreak>anyone using arei ?
<n|Phreak>Nevermind I got it working now
<rlb>Now that we have bytestrings, I wondered about changing the default display format for decoding_error(s) to that. For encodings that are ascii-supersets, that might be a lot easier to follow when trying to figure out what's wrong than the #vu() bytevector representation. Of course that assumes decoding UTF-16 isn't the common case.
<rlb>(i.e. srfi-207 format)
<jcowan>What would people think about writing bytevectors in hex, like #(u)v8(#x41 #x42 ...). Chibi does this with the exception of elements whose value is 0.
<avigatori>has ice-9/psyntax-pp.scm been written by hand?