IRC channel logs


back to list of logs

<guileHacker4000>Hello, is the author of guile-sdl2 in here?
<chitochi>i think it's dthompson? afk right now i think :/
<guileHacker4000>I understand, thank you. I've sent them a message on mastodon, I hope they'll see it.
<apteryx>does scm_internal_catch return either the result of the body, else the result of the handler, if an exception was handled?, the same as in Scheme?
<euouae>yes, exceptions with 'tag' are caught
<apteryx>what does scm_throw do in C, considering there are no exceptions in C ?
<apteryx>phew, I think I'm done figuring out how to replace select with scm_select in guile-udev
<apteryx>entering debugging phase...
<euouae>apteryx: this one might not be documented very well
<euouae>this is what it does, <>
<euouae>I don't think it's thread safe? there is a static throw_var that keeps the state
<euouae>no wait, throw_var is the closure called with (key . args).
<euouae>and later it is initialized with scm_c_define("throw", ...)
<euouae>so I'm actually confused about this part, but I figured I'd explain a bit more by looking at the sources.
<apteryx>that's a good suggestion
<apteryx>would someone know how to fix this conversion problem in C:
<apteryx>I thought SCM_NEWSMOB was obsoleted by scm_new_smob reading NEWS, but I'm not sure what I'm doing
<euouae>something isn't right there apteryx
<euouae>the guile reference I have says the signature takes a void* as second parameter
<euouae>but your compiler claims it's an scm_t_bits
<euouae>see if the latest version of guile documentation + libguile.h headers makes sense: the signatures should agree (on docs and on the header). If not, it's a bug
<euouae>or wait, the description says: " The initial values data, data2, and data3 are of type scm_t_bits; when you want to use them for SCM values, these values need to be converted to a scm_t_bits first by using SCM_UNPACK. "
<euouae>so I think you need to use return scm_new_smob(udev_hwdb_tag, SCM_UNPACK(uhd));
<apteryx>my reference manual (built from source) does say: C Function: SCM scm_new_smob (scm_t_bits tag, void *data)
<apteryx>the value used as data is a struct, not SCM though
<apteryx>a C struct
<apteryx>here's what it looks like by the way:
<dsmith>apteryx, Here is an (old) example. The bot has been using this for about 10 years or so:
<apteryx>dsmith: thanks, I'll check it out
<apteryx>ah, you don't use scm_new_smob() ?
<dsmith>apteryx, It's from 10 years ago...
<dsmith>#define SCM_NEWSMOB(z, tc, data) \
<dsmith> z = scm_new_smob ((tc), (scm_t_bits)(data))
<dsmith>Looks like I am..
<apteryx>NEWS suggest it may be deprecated by scm_new_smob in time
<apteryx>ah! is SCM_NEWSMOB defined to do the necessary cast itself?
<apteryx>I guess that NEWS entry is bogus; that macro seems useful
<tomnor>When recurring like the little schemer, (consing a car to a recursion), is there a limit?
<tomnor>I mean, on the number of "loops"
<apoorv569>Hi, I'm new to lisp languages in general. Recently been using guix and learning guile scheme along side.
<apoorv569>I am writing my shell scripts in guile scheme to practice.
<optimal>tomnor: I haven't read the little schemer, but Guile does have tail call elimination, where recursion in the form of tail calls is optimized so it doesn't "blow the stack".
<apoorv569>I'm curious if it possible to have colored output from shell scripts?
<apoorv569>Like from the `(display` procedure for example.
<optimal>tomnor: The Guile manual has a section on Tail Calls:
<tomnor>optimal: yes, thats great
<tomnor>I have just observed that even if a call is not tail recursive, like (cons something (recur something)).
<tomnor>I run that thousands of loop and no crash
<optimal>apoorv569: guile-lib ( has a `(term ansi-color)' module that can generate the appropriate ANSI color escape sequences for you -
<optimal>apoorv569: You can install guile-lib with guix via: `guix package -i guile-lib`'
<optimal>tomnor: The expression you describe is tail-recrusive
<tomnor>not to my understanding, the last call is cons
<apoorv569>optimal: I see. Thanks for the recommendation.
<optimal>tomnor: The last call is `recur', no?
<tomnor>Maybe my example is poor, but it is an argument to cons, therefore evaluated before the cons is called
<optimal>I may be wrong, but if the last expression is a call to the current procedure (recursion), I believe that suffices for tail call optimizaiton, but I'm not sure.
<tomnor>One can observe the difference using ,trace
<tomnor>my example build up a "stack"
<tomnor>proper tail recursion does not
<optimal>Then I'm wrong.
<tomnor>optimal: maybe, maybe not. check it out, its very interesting to see the trace like that
<tomnor>I seem inclined now to use name lets everywhere to make proper tail recursion
<apoorv569>I noticed that I can't seem to `CTRL + C` cancel out of certain commands I run via the guile script.
<apoorv569>Using the `(system` or `(system* procedure.
<Arsen>ah, welcome to the job control jungle
<Arsen>that happens because of some weird interaction between TTYs and programs ran in them that I don't fully grasp
<Arsen>IIRC the TLDP manpage credentials(7) covers this topic
<Arsen>unfortunately, I don't know of good documentation in a good format on this topic
<tomnor>apoorv569: what about ctrl + d ?
<apoorv569>tomor: Can't do ctrl + d either..
<apoorv569>tomnor: ..