IRC channel logs

<taylan>lechner: I seem to have gotten it to work with this one weird trick https://paste.debian.net/1356514/

<taylan>using "1" as a marker to dispatch to internal "helper" patterns. (it's important that they appear first, and the dispatcher last, otherwise the "1" will match the alist pattern variable)

<taylan>this is one of those annoying things where syntax-case may be overkill, yet syntax-rules is too weak and forces one to use hacks.

<taylan>(unless there's an obvious easy solution I'm missing. didn't write syntax-rules macros in a long time.)

<taylan>ACTION afk

<lechner>sneek / later tell taylan / thanks!

<sneek>Got it.

<cow_2001>okay, now i see that when compiling libgit2 with that experimental sha256 flag it outputs the header files into /include/git2-experimental* but some of them have a line that #includes stuff named git2, not git2-experimental and that is very annoying

<cow_2001>it seems i've come across an alien source repository. now things are beginning to make sense. more later

<cow_2001>if you are confused why pkg-config would not give you flags for a certain library, remember pkg-config --list-all to see if the library you need was given a weird name

<old>cow_2001: don't forger to add pkg-config to your shell for the PKG_CONFIG_PATH to be update on Guix

<old>and yeah, pkg-config is really .. not great. Something so simple, yet it fails at it

<cow_2001>old: i wrote something https://kaka.farm/posts/2025/02/27/work-on-guile-git-and-therefore-gnu-guix-sha256-hash-function-support.html

<lechner>+1 on the pkg-config perception

<cow_2001>lechner: thank you

<old>why is syntax-source sometime returning #f ?

<dthompson>because not all syntax objects have source

<old>that's like saying the color red is red because it is red

<old>To be more specific, what determine wheter a syntax object has a source or not

<dthompson>depends on the situation. a procedural macro could do datum->syntax and not specify source, for example

<dthompson>read-syntax annotates everything with source *if* the input port has file name/line/column info, otherwise it will be #f

<old>dthompson: I see thanks for the explanation

<dthompson>np. generally I'd say that if that syntax object without source was created by a macro then it's a bug in the macro (provided that the input syntax objects had source to begin with)

<dthompson>sourceless syntax objects happen all the time if you are defining things at the repl/using geiser

<old>Well I do have a syntax that create another syntax transformer

<old>this second transformer, I can see its source fine

<old>but the datum that its transform seems to be loosing the source

<old>I'll investigate this

<old>I usually do: (datum->syntax stx my-datum)

<old>I've always assume that my-datum will have the same source as `stx' (syntax-case stx () ...

<dthompson>yes that is true

<old>hmm actually it does not seem to hold all the time

<old>really need to put #:source stx

<old>So this is weird

<old>I get the syntax location whenever my macro is used in a module

<old>but not in a script file

<old>hmm maybe not. I think the problem is from nested syntaxes

<old>hmm I see. I lost the syntax source whenever the expression is in a delay form

<old>or a lambda

<old>Oh god I figured it out. I don't know why, but transforming the syntax to datum then back to syntax was loosing the source location

<old>anyway, fixed

<ArneBab>rlb: your review already prompted a second reviewer to become active ⇒ thank you a lot!

<rlb>glad it helped.

IRC channel logs

2025-02-27.log