<lispmacs[work]>hi, I was just wondering how bad a performance hit is raising an exception, compared to say a procedure call <sneek>lispmacs[work], you have 3 messages! <sneek>lispmacs[work], maximed says: (define-syntax backquote (syntax-rules () ((_ exp) (let-syntax ((exp* (datum->syntax s exp))) exp*)))) (untested) <sneek>lispmacs[work], maximed says: (make-bytevector (backquote (* 100 100))) <sneek>lispmacs[work], maximed says: oops, forgot s is unbound: (define-syntax backquote (syntax-rules () ((_ exp) (let-syntax ((exp* (datum->syntax #'exp exp))) exp*)))) (untested) <lispmacs[work]>sneek tell maximed I am getting an error when I try to use backquote syntax you defined: In procedure syntax-wrap: Wrong type argument in position 1 (expecting syntax object). Maybe show me how you use it in a definition? <sneek>maximed, lispmacs[work] says: I am getting an error when I try to use backquote syntax you defined: In procedure syntax-wrap: Wrong type argument in position 1 (expecting syntax object). Maybe show me how you use it in a definition? <lispmacs[work]>sneek later tell maximed I am getting an error when I try to use backquote syntax you defined: In procedure syntax-wrap: Wrong type argument in position 1 (expecting syntax object). Maybe show me how you use it in a definition? <lispmacs[work]>hi, in the definition of scheme procedure raise-exception (6.11.8.2) there is optional argument #:continuable=#f listed. I got a bit confused by that because it is supposed to be #:continuable?=#f with a question mark. ***chris is now known as Guest7581
***iskarian is now known as Guest5838
<wingo>lloda: i think 3df3ba1a2c956bba122328e1fc4be614171a4f42 is probably an abi breaker; would invalidate files compiled before then with static arrays <lloda>i depend on autocomp so ofc i didn't notice anything <wingo>lloda: i think you can keep the flag change in but just keep the shift to 17 <pkill9>guile-bytestructures reports invalid target for aarch64-linux <pkill9>`guix build --target=aarch64-linux guile-bytestructures` to reproduce <civodul>pkill9: try --target=aarch64-linux-gnu :-) <pkill9>now i need to recompile everything i compiled to aarch64-linux <pkill9>civodul: maybe there should be an error raised then if you dont provide a triplet? <pkill9>i was successfully compiling to aarch64-linux a bunch of stuff <wingo>lloda: sgtm. i want to update match.upstream.scm in that <stis>(bytevector-u64-set! x i (lognot (bytevector-u64-ref x i e) e) is soooo slow <stis>My C version of the loop is 100x faster <civodul>but the thing returned by u64-ref could still be a bignum, no? <stis>The compiler should now to store them unboxed though and keep them fixnum without tagging <stis>which means that it is quit complicated to use lognot in setminus with bitsets <stis>you need to use (logand (2 64-1) (lognot x)) for that <stis>And here ideally the compiler should know what to do <stis>actuaööy even wothout lognot boced values are used <stis>maybe I need to compile with higer OX <stis>(load-compiled "a.scm.go") <stis>does not work, my defined function is not visible why? <stis>it loads the go file though <stis>confusing name of that function <stis>no, even with -O3 the u64 is not unboxed <stis>Oh well I will make some insfrastructure to use C-based code <stis>Anyhow, I got bitvectors on steroids working now, C based speed of many bitwise operations and still friendly to fibers <stis>can use cow and sharing and resizing, <cybersyn>heya guilers! I moved into a friend's house in july who's ip was on a spam blacklist, so its nice to be finally welcome back to IRC <cybersyn>i'm wondering, is there a module for getting Chez-style repl indentation in the guile repl (in the shell, independently of emacs & geiser?) <cybersyn>i'm coming from racket, but after falling in love with guix I've decided to go headfirst into Guile, and I absolutely love it. so I would really like to start operating guix from shell repls, but, while I know it may seem like a superficial complaint, not having the indentation makes it feel a little less smooth; when i'm working in bash, for instance, if i had to count parens, it would feel like a significant hamper on my workflow <cybersyn>also, I've started to get interested in Andy Keep's Nanopass framework, and figured it could be a cool project to port it to guile. has anyone attempted this yet? <clacke>I think not tolerating cycles is a better way to cope with cycles than Python's "appears as if it were working, but then you sort the imports in a seemingly unrelated module and suddenly you have a missing symbol error" <apteryx>Hello Guilers! I'm testing the waters; is a 3.0.8 planned for release soon? <civodul>wingo: speaking of which, i'd like to talk about cross-module inlining <civodul>i had tons of questions/comments when i read your blog post back then, but where did time go? :-) <muradm>hi, in (web client) i'm doing multiple (http-get uri #:port socket #:keep-alive? #t) with same socket. after like 10-20 http requests i'm getting eof-object at some point. <muradm>do i do something wrong? or i have to just reopen socket ? <muradm>wandering if it is me or server or else