<paroneayea>spending all this time in guile over the last year has drastically chaned me as a python programmer! ***dje is now known as xdje
<paroneayea>davexunit: are you writing haunt's docs in haunt :) <daviid`>ACTION is working on guile-squee, git house keeping wrt recent autool chained ... <daviid`>yeah, sorry for the delay, but in an hour or so it's there! <paroneayea>ACTION just popped some apples in the pressure cooker to reduce <paroneayea>nice thing about electric pressure cooker, you don't need to stand around for it <davexunit>I'm gonna make some sort of apple treat tomorrow <davexunit>not sure if pie or dumplings or something else <davexunit>cider, cider donuts, apple pie, apple everything <holomorph>i keep meaning to poke guile's manual, it makes me sad it doesn't have a detailed node listing <paroneayea>initially I thought guile's manual was very short because of that <davexunit>I should be able to easily add texinfo support to haunt <davexunit>I love when the parser is already written for me <davexunit>ACTION pushed texinfo code to haunt master branch. <amz3>Imagine I create a macro as an alias of define using define-syntax/syntax-case <amz3>(define-syntax foobar (lambda (stx) <amz3>This is the skeleton, it's missing the interesting part <amz3>it's missing a period before body, so the pattern transformer should be #`(define (name) . body) <amz3>the problem I have is that I'd like to add something at the end of the body <amz3>to do that, I can use `with-syntax` ? <amz3>well, I replaced `. body` with `body ...` then it's easier for me to add something at the end of generated procedure <amz3>with: body ... (#,(datum->syntax stx 'yield) #,(datum->syntax stx 'eoc)) <amz3>my question is more: how do I manipulate a pattern inside syntax-case using datum->syntax/syntax->datum ? <civodul>nobody here advocating Guile for Emacs? <roelj>When I call a C function that outputs data to a file, it writes the entire expected data, but when I call it from Guile (the same C function exposed with scm_c_define_gsubr) it only outputs a part of the expected data. What could I be doing wrong? <amz3>maybe string->pointer is failing? <amz3>try with strings of the same size, but with simple content <amz3>sorry this is stupid string->pointer is ok <amz3>I remember having a problem with pointer->string not string->pointer <amz3>but the problem was that I wasn't properly packing my c-struct... <roelj>Ok. Let me check on my code.. <roelj>It's really (not) funny. In the function I call another function that fprintf's the data to a file. <roelj>I'm not using any custom types.. <roelj>I figured it out! I need to explicitly fflush(the_file), because fclose() never gets called when the library is loaded into Guile. <rekado>civodul: I would advocate Guile if this thread could be taken seriously. <rekado>David Kastrup actually made the case for Guile already, although it probably was meant more as a lesson from history. <rekado>davexunit: it's "Emacs rewrite in a maintainable language" <civodul>rekado: yeah, emacs-devel sometimes reminds me of bug-hurd in the 2000s: lots of talk, little code <civodul>i followed up via Gmane, but as is often the case, Gmane silently swallowed my messages :-/ <ArneBab>rekado: it sounds like the rewrite idea quickly turned into “move stuff from C to elisp” <ArneBab>which would increase the impact of guile-emacs <rekado>and the backtraces aren't very helpful <davexunit>I hope the people that make the big decisions around there will see the advantages of guile over other options. <ArneBab>I’m not on the list — can someone write about it there? <ArneBab>if you’re on the list, could you answer that this has been done by rebasing elisp on Guile? <davexunit>David Kastrup has pointed this out in the thread <ArneBab>but as far as I saw, there was no one who said “it’s been done” <davexunit>we need to 1) get people on board with Guile as a general approach and 2) write the damn code <ArneBab>reminds me… I wanted to test guile-emacs again <davexunit>bipt has done it all so far. "no programmer is an island" <davexunit>I think someone could *definitely* raise funding for the guile-emacs work if they knew what they were doing. <ArneBab>ACTION has to go looking for jobs which start april 2016 <civodul>OTOH we've already had quite a lot of funding for that project <civodul>it's the social part that's not working as expected <davexunit>the more senior emacs developers and former maintainers have not been convinced. <ArneBab>I would expect this to be a pretty long process <ArneBab>they lose their expertise if they go for that <davexunit>well, not really. the majority of emacs is elisp. <davexunit>we just change part of the C core to use the Guile VM instead of the custom elisp vm. <ArneBab>well, the people working on the C core lose their expertise <ArneBab>and for the others: I fear that guile-emacs needs a one-click install (or needs to be packaged) <ArneBab>we have the same problem for Freenet: The little hurdles count. <ArneBab>it has to be packaged for the distros the maintainers use <civodul>at the same time, there's no Debian package for Guix <ArneBab>if you could tell them: apt-get install guix; guix package -i guile-emacs <rekado>I would love to help on making Guile Emacs a reality but I don't even know where to start. <davexunit>that would be great, but no debian person has so far been interested in making such a package. <ArneBab>for me starting means: I have to fix my guix <rekado>it seems too early to test it, because so many packages seem not to work. <ArneBab>davexunit: I fear that’s not how it works: guix is a potential competitor, so the package for userspace guix likely has to come from guix folks. <rekado>I could live with it being slow if it worked. <davexunit>ArneBab: sure, but I don't know how to create Debian packages because it's so convoluted and terrible. <rekado>I can do RPM but Debian packages are difficult. <ArneBab>with guix I just ran into error while loading shared libraries: libgcrypt.so.11 <davexunit>I use fpm at work as a short-circuit for having to learn it. <ArneBab>is it really that horrible? I thought that they are mostly tarballs… <davexunit>it's not easy, and I also can't find straightforward documentation for how to do it. <davexunit>a guix package is about a dozen lines of Scheme code usually, debian packages are much more complex. <ArneBab>“They get results as opposed to the several failed projects preceding <ArneBab>them. But the main reason is that they don't bother with a coherent <ArneBab>design rather than getting results.” <ArneBab>also regarding guile-emacs: The minimal bar to make this the default version is to make it run on every system which runs Emacs. This is sure to be a long process. <davexunit>I think the only work that would need to be done would be on windows <ArneBab>are the relocatable patches I heard about a few months ago (here) already merged yet (and will they be in 2.0.12)? <rekado>davexunit: and DOS. I think Emacs also runs on DOS. <rekado>(I tried it on FreeDOS but apparantly 8MB are not enough.) <davexunit>I think such a platform could be safely dropped for a large improvement of emacs on GNU systems. <ArneBab>as a note while reading: Emacs is not too slow. <ArneBab>davexunit: not being everywhere means that some people could no longer depend on having Emacs available <ArneBab>do we have numbers which show how many old dos machines are still maintained in industry? <ArneBab>(though it begs the question whether they use old emacs or current emacs) <davexunit>that cannot hold back development of the GNU system, though. <ArneBab>the GNU system also targets rising adoption <ArneBab>(which isn’t to say that this requirement is necessary, just that it’s not quite as simple) <ArneBab>I once read an article about “no one knows what the typical programmer does” <rekado>do the Guile Emacs enthusiasts have a place to discuss bugs and development? <rekado>I doubt emacs-devel would be the right place. <ArneBab>it sounds like guile-emacs could get adoption if it could improve the performance of font-lock (from what people complain about) <ArneBab>^ well, some adoption, because it would be better for some more people <unknown_lamer>part of me just wants to go back to wmaker + cairo-compmgr, if that even works any more... I was seduced by wobbly windows and the pager <unknown_lamer>how hard is it to not rewrite the network and system load monitors to be worse every release :( <unknown_lamer>ArneBab: iirc bipt might be getting some time to work on guile-emacs as part of eir job or something <unknown_lamer>haven't talked to em in a few months though, ey dropped off the face of the planet or something <unknown_lamer>guile-emacs works and works really well too, it can be used as a day to day editor... <unknown_lamer>but bipt literally *refused* to promote it at all, when I worked from /. I tried to get status updates and write articles on it... <unknown_lamer>I think the emacs people haven't been convinced because no one has even tried to convince them <sirgazil>nalaginrut: glad your site feed is working now, because I'm waiting for your guile+robot post :) <amz3>I'm trying to rewrite the GLL in terms of coroutines/generator <amz3>it happens that the left recursive grammars support is implemented in memo-cps <amz3>it looks like memo-cps detects the infinite recursion of a grammar and drop it <amz3>I pretty sure that memo-cps does solve the infinite recursion problem of left recursive grammars <amz3>The problem is that GLL implementation use a strange word to describe the procedure that detect the loop <amz3>never heard of this verb "to subsume" the definition is difficult, and it happens to be also a verb in french <amz3>which is link to philosophy <amz3>for text/sentence segmentation, I really need recursive grammar support because I can't really decide "what's in a quote without using the prior definition of a sentence" <amz3>I mean quote is defined more or less as (define* ((quote) stream cont) (string "\\"") sentence (string "\\"")) <amz3>well this code ^ doesn't run, but that's the idea <amz3>I think I will have try at guile-log again instead...