<delmarre>but running haunt build gets me Error: no code for module <ArneBab_>did you set the GUILE_LOAD_PATH to include the site directory or haunt directory? <delmarre> ("/usr/local/share/guile/site/2.2" "/usr/share/guile/2.2" "/usr/share/guile/site/2.2" "/usr/share/guile/site" "/usr/share/guile") <delmarre>incidentally it all works if i call the ./pre-inst-env first <delmarre>so might just make aliases to do that as a complete hacky solution haha ***bairui_ is now known as bairui
<ArneBab>sneek: later tell delmarre: It needs to be /usr/local/share/guile/2.2/site — not ../guile/site/2.2 <ArneBab>sneek: later tell delmarre: sorry, I got that wrong … site/2.2 is correct <ArneBab>sneek: later tell delmarre: where did make install install to? <sneek>delmarre, you have 3 messages. <sneek>delmarre, ArneBab says: It needs to be /usr/local/share/guile/2.2/site — not ../guile/site/2.2 <sneek>delmarre, ArneBab says: sorry, I got that wrong … site/2.2 is correct <sneek>delmarre, ArneBab says: where did make install install to? <janneke>mwette: rain1 just found that recently (ice-9 textual-ports) from guile-2.2 were introduced in Nyacc <janneke>i think this breaks guile-2.0 compatibility, are you aware of this? <janneke>ACTION had some debugging patches for guile-1.8 and nyacc 0.7x so that nyacc runs on real simple guile (and hence mes scheme) <mwette>janneke: no. I will look into it. <janneke>mwette: thanks. i haven't kept up with Mes<->Nyacc's compatibility recently, working so hard on MesCC only <mwette>janneke: you compiled nyacc to guile-1.8? I can install 1.8 but not confident that one is doable <janneke>it adds r7rs ellipses (for Nyacc), define-syntax-rule, quasisyntax, when, unless, and PEG to guile-1.8 <janneke>i am using (this patched) guile-1.8 to bootstrap Mes' psyntax.pp <janneke>mwette: btw, guile-1.8 has no compiler, so setting load-path is enough ;-) <mwette>janneke: thanks -- looks doable for 1.8; though I may try use-modules + effective-version to get optargs and syncase <janneke>mwette: yes, this made me very happy last year! <janneke>note that this is my enhanced guile-1.8 <mwette>janneke: is the ellipses for the lalr-spec macro? I could eliminate that I think, now that a let is not needed to wrap everything. <janneke>mwette: yes...that would nice for full source bootstrapping... <janneke>...although what would be *really* awesome if we can remove the dependency on syntax-case <mwette>won't happen, as nyacc lalr syntax uses identifier? ; that is what makes the syntax for languages so clean: <mwette>nyacc : ("[" ElementList "," Elision "]" <mwette>lalr: (lbracket ElementList comma Elision rbracket) <janneke>we (from the full source bootstrapping madness) will just have to see <janneke>it is rumoured that syntax-case can be bootstrapped <rain1>maybe we could go over it like ((str "[") (rule ElementList) (str ",") (rule Elision) (str "]")) <sneek>rain1, oriansj says: to message me as I have some important news. <rain1>and modify the macro a bit to do that <janneke>mwette, rain1: maintaining a fork/delta on Nyacc -- just for bootstrapping purposes -- would be an option too <janneke>not the best of options, but keeping Nyacc as clean as possible is also nice <janneke>usually it's not so bad to have some yoghurt software <janneke>managing to bootstrap syntax-case would be another great option <rain1>we will need to think about self extension carefully to get a scheme with macro systems <janneke>lisp-style define-macros are trivial <mwette>actually identifier? is need to distinguish identifiers from symbols, which are not used, usually <janneke>syntax-rules are not trivial, but OK to bootstrap and mes has them <mwette>one would have to declare symbol use (using %token like statement) <janneke>syntax-rules is ~200 lines of lisp style macros <rain1>wow thats awesome, just had a look at mes/syntax.scm <janneke>can you imagine how i felt when that worked? <rain1>yeah must been have been exciting! <janneke>...standing on the shoulders of giants... <rain1>i only have a trivial defmacro in my compiler, first define quasiquote then using it to define cond and similar - i should see if i can pull in that syntax-rules its nice and compact <janneke>rain1: your compiler, is that written in forth? <rain1>no its self hosted on top of a VM implemented in C <rain1>what i had thought before was the forth is very close to this sort of VM (the way it reads in 'words' from a file, load them into a dictionary, execute them with a threaded loop) <mwette>I have simple fix for removing textual-ports. <mwette>`get-string-all' replaced by new proc `read-rest' <mwette>ACTION is working port of nyacc to guile 1.8 <mwette>This 1.8 porting is turning out to be not clean. <dustyweb>mwette: a backwards port of nyacc? why? :O <mwette>dustyweb: janneke using 1.8 - I said I would look. <dustyweb>I wonder if I should try to see if I can use it to parse things in guile-gnutls <janneke>dustyweb, mwette: ouch, hope you didn't misunderstand me there... <amz3>ijp: do you have plans on hacking on guile stuff before next year? maybe guile-js stuff? <amz3>ijp: I would be glad if you give me feedback on my commits <dustyweb>I'm having a hard time figuring out by using it what useful data I can get out tho <janneke>i backported things needed for Nyacc to guile-1.8, to help boot Nyacc on Mes <janneke>currently, I use Nyacc on guile-2.2, and on Mes <janneke>when I have a problem with Nyacc on Mes, debugging with Guile-1.8 helps because Guile-1.8 and Mes are closer <janneke>however, i *also* use this patched Guile-1.8 to bootstrap psyntax.pp for Mes <mwette>janneke: OK then. I did remove dependence on ellipsis and srfi-43, but the parser is not working. <janneke>this psyntax.pp I use in Mes to run Nyacc... <janneke>mwette: still, it would be very helpful for bootstrapping if those were not needed <janneke>not urgent, but for bootstrapping it's nice to have less macro-fu dependencies <janneke>i spent quite some time on removing ellipsis from Nyacc and then opted to backport ellipsis to Guile-1.8 and then to Mes <mwette>janneke: wow -- nice effort there <janneke>mwette: eh, i spent time and failed -- then took the "let's add and use ellipsis route" <janneke>so i spent time on "trying to remove" <mwette>I have removed the use of ellipsis, but getting error for 1.8 on simple spec <janneke>but getting ellipsis to work on guile-1.8 was kind of cool :-) <amz3>dustyweb: héllo! What is the guile web tutorial you are talking about in the ML? <amz3>dustyweb: direct link? :) <janneke>mwette: hmm, i had silly things with port encodings but that was guile-2.0 i think <janneke>mwette: i think that guile-1.8 does not have `when'? <mwette>I ran into unless. I added compat18 module which provides unless via syntax-rules <janneke>mwette: or does not support reading #\\b, does not have #\\alarm, #\\vtab <mwette>guile18 also does not have srfi-43 <janneke>indeed, i backported that for guile-1.8 too <mwette>also, make-fluid does not take an argument. The value defaults to #f but I was using (make-fluid #f) anyhow so I changed from (make-fluid #f) to (make-fluid) <mwette>also syntax case reader macros aren't there (i.e., can't use #' must use (syntax)) <janneke>yeah, backported those too to guile18 ... <daviid>lloda: I don't remember the name and location of your CBLAS related project and repo? <civodul>dustyweb: i haven't looked closely at libtasn1, but it's what gnutls uses <civodul>though parsing ASN.1 in C is debatable :-) <dustyweb>civodul: what's the name of the scheme library that did all the crypto stuff? I couldn't remember it today.. <civodul>perhaps CHICKEN, Racket, or Bigloo have an ASN.1 parser too