<rlb>If civodul responded wrt the i386/i686 numbers.test failures, I missed it (got disconnected), but I have a plausible patch now that makes just those tests unresolved there for the time being, and I'll likely replace the blanket test suppression in the most recent NMU with that soon. <sneek>Welcome back rlb, you have 1 message! <rlb>(and almost certainly better than ignoring all test failures, which is what the nmu did) <rlb>Understandable, since marking them unresolved is not entirely trivial if you're not used to guile, etc. ***catonano_ is now known as catonano
<rlb>I wondered if the examples/Makefile.am really means to fully expand all the vars based on the configure time values. i.e. if it's intended for use with GNU make, then I could easily see wanting to leave $(CC) as $(CC), not as /usr/exciting/bin/gcc-x86... <rlb>i.e. Fully expanding creates an example/Makefile that won't work at all if the system that the package is installed on is different (say wrt usrmerge) from the build system. <rlb>Course maybe it's only intended to "give the flavor" with respect to what's required... ***sneek_ is now known as sneek
***apteryx_ is now known as apteryx
<leoprikler>After reading a bunch of stuff on usrmerge: "There is no way to reliably bring up a modern system with an empty /usr". Uhm, yes, there is? You simply need to put your stuff into /{gnu,nix}/store ***jonsger1 is now known as jonsger
<str1ngs>leoprikler: or you can just init=/opt/emacs/bin/emacs :P ***jonsger1 is now known as jonsger
<stis>Ahh got python-on-guile working for 3.0.4. I willnot support the earlier 3.x.x versions <ruffni>how can i read the output of (system "foo") call (for example as string into a variable)? i figure it's some kind of port-redirection? <pinoaffe>ruffni: usual port redirection directives don't work with system, you can do something along the lines of (let* ((pipe (open-pipe* OPEN_READ "echo foo")) (string (read-line pipe))) (close-pipe pipe) string) <ruffni>sooo, did you manage to find a solution? in your example above i can't see where the (system) call would fit in.. <pinoaffe>ruffni: rather than use the system function, you use the open-pipe or open-pipe* command <pinoaffe>as far as I can tell, there's no way to redirect the output of system <manumanumanu>If I define a procedure in _the same module_ in an (eval-when (expand) ...) syntax-local-binding can resolve it! I am a very happy man. <manumanumanu>civodul: this saves me sooo much work. I thought I would have to write a utility to fully expand all macros within a syntax object <manumanumanu>whereas I can now cheat and just have people not define loop clauses for my looping facility as procedures. <manumanumanu>The option would be to write CPS syntax-rules macros, which is 1. bad for error reporting and 2. insane. <manumanumanu>which is fine for foof-loop, but I am adding subloops and explicit accumulators. The amount of state I would have to throw around in the macros would have been unbearable. my first semi-working draft had this (loop-next ((lets ...) . lets-rest) ((fors ...) . fors-rest) ETC ETC. lets, fors whens, breaks, finals, accs, withs. <manumanumanu>civodul: oh yes! That's where i first got the idea. first response was "nah. not practical", then I found out that both alex shinn and taylor campbell did it, and it was the only way to do it in a low level macro facility. I almost cried. <manumanumanu>but syntax-local-binding together with some serious parameter abuse lets me hide most complexity apart from 2 places, which is manageable. <manumanumanu>the macro for just parsing the clauses had over 20 clauses itself, and that made use of quite a bit of auxiliary macros to handle complexity. I ended up being scared to touch it, even though it was just barely in a usable state. <manumanumanu>the racket for loops has a much simpler iteration protocol. You just leave the hard part to the inliner. But then you lose the ability to do direct transformations. In racket you can't do (for/stream ((a 0 b) (b 1 (+ a b)) a) to get a stream of fibonacci numbers. You will have to do it as a fold. <stis>I think I found a serius bug in guile 3.0.4, not sure but the result is strange <stis>I want to investigat more but inlining of funcitons may be corrupt