IRC channel logs

2015-10-11.log

back to list of logs

<abunai>hi, I'm having a bit of trouble emerging dev-scheme/guile on gentoo
<abunai>from what I can tell, it's attempting to make directories and write files outside of the portage sandbox
<abunai> https://bpaste.net/show/ac34778eeb27
<abunai> https://bpaste.net/show/e7e4a278f8bf
<abunai> https://bpaste.net/show/64f97c806a0b
<abunai> https://bpaste.net/show/1dcba56c0d37
<amz3>abunai: guile-1.8.8 is the old stable
<amz3>IIRC the stable guile 2.1 on gentoo works
<amz3>IIRC, it doesn't exists in gentoo, what I did was copy the guile-1.8.8 recipe and update it for 2.1
<amz3>1.8 is too old, you'd better try to get 2.1 working
<abunai>I'll see if I can get 2.1 working
<abunai>the emacs stuff in the portage tree should link fine gainst 2.1 right?
<amz3>emacs stuff?
<amz3>abunai: ?
<amz3>which emacs stuff
<amz3>abunai: look in overlays to see if there is any guile-2.1 ebuild
<abunai>that's actually what I'm doing atm
<abunai>emacs with the use flags I've selected requires guile
<amz3>I'm not sure why it requires guile, i'll check
<amz3>abunai: actually it's 2.0.11 not 2.1 sorry
<amz3>it's in portage
<amz3> https://packages.gentoo.org/packages/dev-scheme/guile
<amz3>nice gui btw
<abunai>I checked that already
<amz3>it doesn't work?
<abunai>but there is an explicit message portage gave me that said emacs does not build against it
<abunai>didn't try it though
<amz3>which emacs ebuild to use? because emacs-24.5 doesn't reference guile in its useflags
<amz3> https://packages.gentoo.org/packages/app-editors/emacs
<amz3>you talk about guile-emacs?
<abunai>one sec, I'll paste some emerge output and the use flags I'm using
<amz3>I hope one day guix will have as many package as gentoo :)
<amz3>ArneBab_: I agree with you regarding definition expression, what I do is put every definition at the top of the previous definition expression
<abunai> https://bpaste.net/show/b38d897c877d
<amz3>ArneBab_: from a beginner point of view it's misleading to not be able to use define anyware
<abunai>that's emerge trying to pull guile
<abunai>for emacs
<abunai>and although I can probably just remove the offending use flag(s), wouldn't it be better to be able to emerge guile? ^^
<amz3>I'm not sure but it's might be gnutls that pulls guile, even if guile useflag is off
<amz3>abunai: it's not always a good to install every dependency, maybe guile is only used to script gnutls
<amz3>that's the point of gentoo :)
<abunai>yep :)
<abunai>but is gnutls needed for emacs package repos?
<abunai>or maybe it was my password manager thing
<abunai>pretty sure I needed it for something
<amz3>gnutls only accept guile-1.8
<amz3>but it should be possible to install gnutls without guile
<amz3>abunai: I think it's regression or something, you should report that to #gentoo or fill a bug
<amz3>but something strange is happening
<abunai>I already gave a full report in #gentoo and made a forum post
<amz3>net-libs/gnutls-3.3.17.1::gentoo USE="crywrap cxx nls openssl zli
<abunai>no reply in either, but the forum post might get something in a few days
<amz3>b -dane -doc -examples -guile
<amz3>this explicit that guile disabled
<abunai>I wanted to ask around before going straight to filing a bug report
<abunai>I'll give that a try and see if it keeps it from pulling guile
<amz3>I'm mean, that it should not pull guile since the useflag is off
<amz3>sorry I can't test myself, I don't use gentoo anymore
<amz3>maybe try to update portage and see whether something changed
<abunai>luckily I'm here to do so
<abunai>I tried updating portage, overlays, everything
<amz3>I installed multiple times gentoo and emacs on gentoo never saw that issue
<abunai>I even tried letting the guile ebuild do its thing outside of the sandbox
<amz3>that is bug for sure
<abunai>and it finished, but didn't work
<abunai>well that's strange
<abunai>I tried emerging gnutls with those flags
<abunai>and it is still trying to give me guile-1.8.8
<amz3>which gnutls version it is ?
<abunai>3.3.17.1
<abunai> https://bpaste.net/show/a29d8af7e1c8
<amz3>I found this in the ebuild: RDPEND="guile? ( >=dev-scheme/guile-1.8:*[networking] )"
<amz3>do you know what [networking] means?
<amz3>IIUC if USE="-guile" it should silence the dependency
<abunai>yeah, I thought it should too
<amz3>this is very strange, because the useflag is considered off, but it still pull guile
<amz3>I don't recall gentoo doing such thing, you always need to set use flags manually
<amz3>I mean gentoo never set useflags automatically without user input
<amz3>so maybe, but unlikely, the use flag is set in the useflags file (I don't remember the name)
<amz3>but that isn't even an option, since if it was the case the USE env var would override it i guess...
<abunai>it should, yes
<amz3>you surely need to fill a bug for guile-1.8 that doesn't compile
<amz3>and gnutls
<abunai>alrighty
<abunai>I'm just surprised I don't see anything about this already
<amz3>ah
<abunai>1.8.8 is pretty old and used by a good number of packages that are common
<amz3>do you use multilib?
<abunai>yes
<amz3>because the configure flag is enabled with "$(multilib_native_use_enable guile)"
<abunai>actually
<abunai>I don't think I have multilib set up on this install
<abunai>just checked
<amz3>yes, anyway this is the configure step
<amz3>that's not what pulls the dependency
<amz3>in theory RDEPEND does
<amz3>and DEPEND
<amz3>what you can do, is ask someone else to try to install emacs/gnutls and see what happens
<amz3>or build a vm to do that yourself
<amz3>you probably can do it from the livecd
<abunai>yeah I can boot into my liveusb
<abunai>actually I can probably just mount and chroot to try
<abunai>didn't think about that
<abunai>well that's painful
<abunai>I only have minimal CDs around on read-only media
<abunai>anyway, I'll wait to see if the forum post gets a response or if someone cares to have a look on #gentoo tomorrow
<abunai>thanks for the help and advice amz3
***mario-goulart` is now known as mario-goulart
<davexunit>well, here I am, inventing my own document language ;)
<paroneayea>davexunit: heh :)
<davexunit>I'm finding Skribe to be unideal, though I will add it to Haunt for rekado. :)
<paroneayea>I was just listening to a podcast where they were talking about racket's documentation language. It made it sound pretty good.
<davexunit>they have skribble and pollen
<davexunit>I don't know if I should even try, I'd like something like Skribe, but I'd like literal text to be the default, not something you wrap in []
<davexunit>paragraphs are newline separated
<paroneayea>sounds interesting
<paroneayea>might also be useful for, say, when you need to render plaintext emails and need a templating language
<paroneayea>and etc
<davexunit>code blocks and block quotes are indented
<davexunit>(not sure how to differentiate them
<davexunit>)
<paroneayea>I'm sold on sxml for xml templating
<davexunit>and you can insert arbitrary Scheme code by unquoting
<paroneayea>but plaintext obviously needs something else
<davexunit>yeah
<davexunit>agreed
<davexunit>I want something that is a plain-text format + the ability to insert Scheme expressions anywhere
<paroneayea>that sounds great.
<davexunit>This is a test ,(+ 1 2 3) of davexunit's document format
<davexunit>would evaluate to
<davexunit>'("This is a test " 6 " of davexunit's document format)
<paroneayea>oic
<paroneayea>each paragraph is a string in the list?
<davexunit>or rather
<davexunit>'(p "This is a test " 6 " of davexunit's document format)
<paroneayea>oh wait
<paroneayea>I see.
<paroneayea>yes
<paroneayea>got it
<paroneayea>davexunit: interesting idea!
<davexunit>if the translation is straight to SXML
<davexunit>which seems reasonable enough
<davexunit>because it will be a subset of HTML5 and easy to convert
<paroneayea>davexunit: I wonder if you can put it as a layer on top of scheme on the guile compiler tower! ;)
<davexunit>I would implement it as a reader
<davexunit>like skribe is
<davexunit>using guile-reader
<paroneayea>I suppose that makes sense
<paroneayea>I haven't looked at guile-reader but it sounds compelling
<davexunit>now actually making the reader do the right thing... I have no idea.
<paroneayea>:)
<paroneayea>okay, well I have my own thing to write... by the end of the day I hope to have most of this python activitystreams library in place.
<paroneayea>gonna disconnect from irc for a bit
<davexunit>good luck!
<paroneayea>davexunit: sounds like fun and look forward to hearing more
<paroneayea>you too!
<davexunit>I'm gonna disconnect soon too and see my family
<davexunit>but I'll hack until then
<davexunit>later!
<amz3>I coded the thing at least twice, I'm finally happy with the code
<amz3>the new api allows to define declartively a table and its indices
<amz3>so say (we all) that we have one table 'messages' and two indices 'user' and 'published-at'
<amz3>one can use a list to define that, in single block of code
<amz3>previously it was spread over two proc calls which use the same information
<amz3>this was the 'what?'
<amz3>the 'how?' I am very happy
<amz3>I defined my new procedure as a star procedure: session-create* where session-create is the raw procedure that creates tables
<amz3>when I started session-create* was calling deep inside the list mangling code session-create
<amz3>instead what I do now, I have a basically a "specification->session-create-arguments" procedure
<amz3>and I do (apply session-create (spec->args spec)) in session-create*
<amz3>it's basicaly the same code, except I don't do a tail call
<amz3>and I can test the code
<amz3>more easily that is
<amz3>*some kind* of tail call
<amz3>in fact it makes possible unit testing. I can test session-create* helper procedures without testing at the same time session-create
<amz3>previously I had a (make-table-factory spec) which returns (make-table session) procedure which was calling session-create
<amz3>!
<amz3>CPython is moving to gitlab
<amz3>instead of the python kallithea because of lake of hands
<amz3>lack
<ArneBab_>there’s now an ebuild version of py2guile: https://www.epubli.de/shop/buch/47802
<ArneBab_>s/ebuild/ebook/
<ArneBab_>also I added an update info for wisp 0.9.0 which no longer depends on Python for bootstrapping: http://draketo.de/english/wisp
<paroneayea>hi
<paroneayea>heya sirgazil
<paroneayea>how goes?
<sirgazil>Hey, paroneayea :)
<sirgazil>Everything fine, and you?
<paroneayea>sirgazil: good; making good progress on a python library for advancing federation standards work
<sirgazil>great
<sirgazil>I'm experiencing a courious thing generating translation cataloques from guile files with xgettext
<sirgazil>For some reason, strings marked for translation with (gettext "string") are extracted, but if I use the shorthand (_ "string"), they don't.
<paroneayea>sirgazil: _ is often used for "not needed" in guile but I think that's convention rather than actual code...
<paroneayea>er
<paroneayea>rather than actual syntactic stuff
<paroneayea>so it's probably not that
<paroneayea>thinking out loud, sorry!
<amz3>héllo sirgazil
<sirgazil>Ça va, amz3?
<amz3>oui :)
<amz3>and you ?
<sirgazil>Ça va :)
<amz3>I wanted to ask about the guile introduction in the new website
<sirgazil>Sure
<amz3>do you have a particular plan?
<sirgazil>Not really.
<sirgazil>amz3: I just know I want one, but with my current skills, I don't think I can contribute much to it.
<amz3>well, I've read that the best introduction tutorials are written by people that are not experienced
<amz3>I'm not sure, how that is true
<paroneayea>my first contribution to Hy was a tutorial
<paroneayea>and that was trying to make sure that I myself understood the language
<paroneayea>most of that tutorial is still there!
<paroneayea>in the docs!
<amz3>:)
<amz3>I'm looking at it
<amz3>the only tutorials about lisp I know are books
<amz3>it's good, I will take as skeleton for guile tutorial
<sirgazil>In any case I would like to help with the introductory material in any way I can. I'll think about it.
<paroneayea>\\o/
<rekado>sneek: later tell davexunit The proposed skribe document syntax looks good to me! Re indentation: I was using a mix of scheme-mode and manual indentation for my blog posts.
<sneek>Will do.
<rekado>sneek: botsnack
<sneek>:)
<ArneBab_>taylan: thank you for your answer in guile-devel!
<ArneBab_>I answered David, because I had already wrote most of my reply to his off-list answer when I saw your message
<taylan>happy to help :)
<taylan>internal definitions are pretty tricky, I also find myself thinking of possible semantics for intervined definitions and expressions but all the solutions seem non-satisfactory
<ArneBab_>yes