logs archiveIRC Archive / Freenode / #emacs / 2010 / January / 8 / 1
Intensity
quotemstr: Ok, so I show #14225 0x080e5862 in Fgarbage_collect () at alloc.c:5083. Is that the point where I do one more "up" followed by "ret"? Or do I do "ret" here?
quotemstr
Intensity: Ffoo functions are the C embodiments of builtin Lisp subr, so they're good places to return from.
Intensity
quotemstr: Ok, then should I do one more up so that gdb's last stack print is Fruncall_run_hook_with_args instead of Fgarbage_collect? Here I have the aim not only of trying to recover the session, but hopefully providing some more information to developers.
quotemstr
Yes.
Intensity
quotemstr: Well I don't want to delve too deep into this in the main channel, but after that I got: #0 signal_before_change (start_int=31, end_int=32, preserve_ptr=0xbfa865b0) at insdel.c:2199.
quotemstr
Intensity: Now type 'continue'
And wave a dead chicken in the air while invoking ancient gods.
Intensity
quotemstr: Program received signal SIGSEGV, Segmentation fault. search_buffer (string=<value optimized out>, pos=<value optimized out>, pos_byte=30, lim=41, lim_byte=41, n=1, RE=0, trt=138590812, inverse_trt=139264924, posix=0) at search.c:2088
quotemstr
Intensity: Ahwell. :(
Intensity
quotemstr: I didn't have a chicken to hand... Is this the final word for this session then?
         

quotemstr
Intensity: Yeah. You can't recover from that.
Intensity
quotemstr: Ok. Which build options are best for me to enable in emacs so that I'm likely to get more detailed information here in the future without significantly slowing down emacs?
quotemstr
Intensity: Nothing, really. The most helpful thing would be a reproducible testcase.
Geralt
quotemstr: I tried it with the variable first, but it's not working, here's my test-code http://zlin.dk/p/?MWY4NTBi can you see anything wrong with that?
Intensity
quotemstr: Ok. Are there any options in header files that I can beef up to avoid even encountering a problem like this?
quotemstr
Intensity: No, but you can try manually running (garbage-collect) every so often.
Intensity: And keeping track of memory-use-counts.
Geralt: You need the *value* of comment, start.
Geralt: But for lisp, it's so simple that you might as well just use ';'.
Intensity
quotemstr: Ok. Is Emacs CVS generally moving toward more stability right now (like focusing on bugfixes only)?
quotemstr
Intensity: Couldn't tell you that, sorry. Generally Emacs HEAD is pretty stable for normal use, however. But if you're only restarting every few months...
Intensity: Also, Emacs switched to bzr. It's CVS no longer. :-)
Geralt
quotemstr: thanks, with ";" it works, but how can I get the value of comment-start?
Intensity
quotemstr: I'd just prefer to use the most stable version. I thought momentarialy that I might find a more stable version than the latest release.
quotemstr
Geralt: `((,comment-start ...))
Geralt: Of course, comment-start needs to have the correct value at the time you make the call, meaning you need to make it from a mode hook.
Intensity: The latest release is going to be the most stable.
Geralt
quotemstr: thanks, I'm intending to call it in a hook, that one is just for testing at the moment
Intensity
quotemstr: Ok, while I'm at it, I'd like to somehow periodically write memory data somewhere. I have a periodic command I attempt to enter into emacs but it doesen't seem to periodically run as intended: (run-with-idle-timer 3600 t (lambda () (mapc (lambda (buf) (set-buffer buf) (when (eq major-mode 'erc-mode) (erc-save-buffer-in-logs buf))) (buffer-list))))
quotemstr: Does everything about that look right? Perhaps I'm not actually considered to be "idle" from emacs' perspective, and I could run it with (run-with-timer ...) (if that exists). Maybe I can write a similar line to dump memory data periodically too.
quotemstr
Intensity: The code looks correct. You can write memory information using a temporary buffer and append-to-file.
ggole
Intensity: that will only run when emacs is idle for an entire hour
konr
The region is between the mark and the point, right?
Intensity
quotemstr: Ok. I don't even know enough lisp to be able to do that, but perhaps I could work it out on my own.
ggole: Ok, is there an alternative that will run that unconditionally? Is the only criteria for emacs idleness whether I've entered a keystroke recently?
ggole
Uh, I don't remember the exact circumstances
There's an entry in the info files iirc
Intensity
Hmm. While I'm at it, I've never been able to run the Google client code found in emacspeak. I continue to get lines like "process speaker <n>" not running. I don't know if there's a kind of no-op server I can run that will satisfy emacspeak's requirements to merely let me load and use some of the Google client code.
And, more generally, is there a company that is strongly concerned with the stability of emacs, such that I may copy them on a bug report? Any paid developers who are likely to want to look into this problem in detail?
quotemstr
Intensity: http://paste.lisp.org/display/93119
Intensity: You might get a better response on emacs-devel if you offer a bug bounty. :-)
Intensity: But AFAIK, there are no professional paid support services.
         

dunnen
ouch
woops, wrong window
quotemstr
,salespitch
fsbot
We aren't gonna lie. Emacs sucks. Some of us tolerate it, but we can't tell you if YOU'LL be able to. Try it and make up your own mind.
ggole
I think a good report of a serious bug is very likely to see attention
Intensity
quotemstr: Cool, thanks for drafting that. Also, thanks for the information on midnight-mode.
quotemstr
Intensity: You're welcome. I hope this bug receives the attention it deserves. GC bugs are serious.
ggole
The question is not so much whether the devs will care as whether you can give them a good reproduction of the bug
quotemstr
Intensity: Again, if you can reliable reproduce the problem (preferably on timescales shorter than months), that would be very helpful.
Intensity
ggole: Yeah I see what you mean. Hopefully the line numbers that are provided in the backtrace may be helpful, but I guess that's the most they'll be getting. It's possible that if someone generally uses up a lot of memory then it wouldn't take so long to be able to reproduce. Clearly next time I won't use a gnus dribble file because that's the only large buffer that I was "editing".
quotemstr
s/reliable/reliably/
Intensity: Also, in the bug report, mention the *precise* version you compiled; otherwise, the line numbers are less useful.
Intensity
quotemstr: I'm not too far from offering a bug bounty, but to be fair, I've encountered so many bugs. I "live in fear" when I use viper, for example. It doesn't seem to work properly when on a tty. Exiting out of <I>nsert mode followed by another command mode keypress does crazy things that are interpreted like M-y (where y is some keystroke--such as 'k').
quotemstr
Intensity: You're at least reporting the bugs, right?
Intensity: I don't use viper so I can't speak to that particular issue.
konr
What's the point of using pretty print and not using the *Pp* buffer?
Intensity
quotemstr: I've run out of time even to report the viper bug, but I've mentioned it on the channel.
konr
Intensity: have you used vimpulse?
quotemstr
Intensity: Things get lost in the channel, regrettably. M-x report-emacs-bug at least puts it in a fixed medium.
Intensity
quotemstr: It's pretty important to me to get fixed, but it's general enough that I'd think someone else must have noticed ("Viper doesn't work properly in a tty").
konr: Nope, I've not heard of that, but I'm checking now.
e1f
"It was Comdex 2001...Gates predicted the tablet would become the most popular form of PC within five years."
konr
Intensity: Here is the updated version: http://www.assembla.com/spaces/vimpulse/
bpalmer
e1f: instead, it was the cell phone.
tablets were too big.
Sketch
s/were/are/
e1f
and laptops
Intensity
konr: Thanks for letting me know about it. Seems that it works alongside viper.
e1f
http://www.theregister.co.uk/2010/01/07/microsoft_tablet_pc_slate/
Geralt
quotemstr: I just noticed that for C comment-start is set to "/ *" but then in the statement "int a = 3 / 5 * TODO;" everything following the / is seen as the comment. I can live with that, but it looks strange to me that Emacs uses such a regexp for comment-start
quotemstr
Intensity: I just tried to reproduce your viper bug. It works fine here.
konr
Intensity: yes! and upgrades its functionalities... take a look at viper-in-other-modes too... It's a little buggy, but I'm currently working on that
e1f
and apple is not going to call it a tablet because of the connotations
it will be a new name
quotemstr
Geralt: comment-start isn't actually used for comment detection. Syntax tables are there for that, and they're more complicated and subtle.
konr
Intensity: http://www.assembla.com/spaces/viper-in-more-modes/new_items
bpalmer
e1f: a newton?
quotemstr
Geralt: comment-start is used mainly for comment-region and a few quick and dirty hacks.
e1f
mabe isaac
quotemstr
e1f: etch-a-sketch?
e1f
or iSaac
Geralt
quotemstr: I see
ggole
"/ *"? Not "/*"?
Intensity
quotemstr: If I rapidly try to exit and enter in and out of <I> and <V> modes, then an unexpected command will be run instead. For example, start in <V> mode over some text, and keep typing "ESC a" rapidly in a cycle. The cursor should ideally stay put, but eventually the sequence is interpreted as a command.
Geralt
ggole: typo, it's "/* "
Intensity
konr: That's great. Do you know if there's a stronger way to report my viper difficult(y|ies) apart from a standard emacs bug report? Perhaps emailing the author of viper? Or do you take interest in this bug? Unfortunately, because it exists, I have to live in fear.
ggole
Intensity: if you don't get satisfaction from a bug report, an email to emacs-devel or the viper maintainer might be worth a try
quotemstr
Intensity: That's odd. I cant' even reliably exit from <I> mode.
konr
Intensity: well, there is a list of vi-related projects which has the authors of vimpulse: implementations-list@lists.ourproject.org . Maybe the autor of viper is there, although I was advised to do a bug report when I wanted to suggest new features to viper
quotemstr
Intensity: I think it's the overloading of ESC as 'meta' that's causing the issue. The problem seems to be the relative speed.
Intensity: Here, 'ESC a' 'ESC ESC a', etc. cause problems, but ESC <small delay> a works fine.
Intensity
quotemstr: I've come across *very* strange behaviour. Sometimes I'll type ESC and instead of exiting out of <I>, I'll see an "ESC" at the foot of emacs. Then even several ESC's in a row aren't seen properly. I simply don't use function keys so I have no need for translations of those sequences from the terminal. So that's why I have viper-fast-keyseq-timeout viper-ESC-keyseq-timeout both set to zero. I suppose that's the right thing to do i
quotemstr: Yeah, I don't want to have to wait around. Especially if there is a small delay to the server anyhow, then the delay becomes too inconvenient. And in any case one of the other bugs is that even if I press ESC in <I> mode and wait, Viper doesn't actually exit out of <I>nsert mode.
quotemstr
Intensity: Yeah, post to emacs-devel about that. It's almost certainly the ESC-as-meta handling in the tty layer that's causing the problem.
Intensity
Intensity: And at one point the ESC's at the foot of emacs (below the modeline) just kept piling themselves up on their own, and kept cycling even though I wasn't entering any new keystrokes.
quotemstr: Ok, I'll do that. Perhaps a configuration option could be offered to simplify viper's functionality here.
quotemstr
As I said, I don't use viper, so my ability to comment here is limited.
Now, if you'd found a bug in the Javascript support, I'd be much more proactive. :-)
Intensity
quotemstr: Ok. I have some bug reporting and rebuilding to do.
In the meantime, the last thing I'd like to ask about (since it will affect the new emacs session I'll run) is whether there is a way to avoid the "Process speaker<n> not running" errors and setup a null emacspeak server. I'd like to be able to use Google Client services as part of that code.
quotemstr
I don't use emacsspeak either, sorry.
sshelagh
you don't need emacsspeak to use the g-client stuff
I have that set up without emacsspeak and it works
Intensity
sshelagh: Ok, so there aren't referring calls to the other parts of the emacspeak within the g-client code? What's a lightweight way to get that working?
sshelagh
I think I 'll have a look
(load-library "g")
;;; greader-reading-list
;;; greader-subscribe-feed
;;; greader-find-feed
Is what I have in my .emacs
Intensity
Ok. Thanks to all for your support here. Emacs is a great technology in spirit; I'd like to see so many bugs resolved. Perhaps an enterprise Linux company that puts out distributions might decide that it's worthwhile to offer a major product suite in emacs, and thus may offer resources to fix more bugs. For now the best option I see is to use not only a stable version (restricted to >= 23) but a release point that only receives stable update
sshelagh
there may be some gmail stuff elsewhere I've forgotten about.
But I basically followed the instructions in the package
e1f
emacs is a niche product
bpalmer
Intensity: you mean a company might decide to write a C++ compiler/IDE suite, and offer to pay an emacs developer full time, and significantly improve the product, and run into the brick wall that is RMS?
e1f
99% of the world hasn't heard of emacs
Intensity
sshelagh: Ok. Hmm. So all you basically do is the load-library? What do those listed functions refer to after the comment symbol? And did you unpack g-client from the website separately? Or use the latest one bundled up with emacspeak?
e1f
and the other 1% don't care about those 99%
sshelagh
I got a separate package
the comments are a reminder of the functions I need to call to get greader running.
quotemstr
bpalmer: RMS doesn't have much of an influence these days, it seems.
Intensity
bpalmer: I'm not sure. I'm trying to create a realistic framework to imagine how emacs may improve its stability at some point (in particular with the packages) but I don't quite see how or when yet.
sshelagh
I don't use it that often.
bpalmer
Intensity: sorry, I was quoting from history.
That was XEmacs's origin story.
Intensity
sshelagh: Ok, I think that the separate package that's listed is about two years old, whereas the one bundled with emacspeak is updated recently.
sshelagh
There are supposed to be ways to use the google calendar too but I'n not interested in those
bpalmer
quotemstr: bzr
e1f
http://www.appleinsider.com/articles/10/01/07/dell_previews_5_inch_tablet_concept_akin_to_bigger_ipod_touch.html
quotemstr
bpalmer: Isn't much of an influence over emacs proper, but rather its development.
Intensity
sshelagh: Perhaps I could delve into using Google Calendar (or orgmode for that matter) but the main thing I'd like to be able to use is the ability to read/write Google Documents, which I believe is available in g-client.
sshelagh
that may be possible.
They use the odf format.
But even though it is part of the emacsspeak package it may still be possible to load it without setting up emacsspeak
jlf
emacs is just a fad. you'll see.
bpalmer
jlf: who would want to run one on tablet computers, after all?
jlf
ya i know
konr
What else pretty printing does besides indenting?
rgr
How could I get emacs to load a file after it has drawn its very first frame (its an encrypted file which cause epa to prompt for a password)?
HFSPLUS
VIM is better
rgr
you're right.
damd
indeed
ggole
HFSPLUS: go away
HFSPLUS
cuz u know im right
rgr
I think he makes a valid point.
jlf
ggole: no, let him speak!
Ober
hmm so C-s is search what is the shortcut for the next reference?
konr
of course
r4yner
HFSPLUS: there is an emacs package to emulate vim. qed
damd
i agree with HFSPLUS though, vim is much better. one tool for one job, one job do it well, all that.
jlf
vim can also double as a roman numeral. can emacs???
fsbot
mu!
rgr
thats always the nail in the coffin ;
ggole
Ober: C-s again
Ober
reprompts me
rgr
no it doesnt
damd
Ober: do something like: C-s a C-s C-s C-s
that should give you the idea
Ober
thanks
r4yner
damd: i used to be a vim user and still use it from time to time, but orgmode and auctex are essential to me
jordanb_
@#emacs lets wrt our irc msgs like like #twitter lol A++ http://bit.ly/5soFV
rgr
auctex is overrated imo
damd
r4yner: i'm just "re-trolling" him :)
or counter-trolling rather
legumbre
RT @jordanb @#emacs lets wrt our irc msgs like like #twitter lol A++
jordanb_
RT @rgr $auctex is overrated Yah me too!!!11
damd
oh so *that's* what RT means... i finally get it.
rgr
?
damd
"reply to", isn't it?
jordanb_
"re-tweet"
damd
damn...
rgr
theres an emacs tweet page???
fsbot
mu!
jordanb_
It's when your ability to communicate has decayed so much you can't even bust out your own semi-literate text message so you have to rip off someone else's drivel.
bpalmer
rgr: yes.
rgr
or does it crunch up text to kiddy talk?
jlf
rudybot: quote
rudybot
jlf: Let's take the worst fu*king part of a computer and make that the interface to everything.
rgr
I dont get it. sorry.
jlf
i like how rudybot retweets jordanb
jordanb_
Me too.
bpalmer
rgr: what's there to get? Someone created an emacs twitter account, and tweets from it.
jordanb_
I bet it's sacha.
Or e1f.
rgr
so whats the RT stuff???
fsbot
mu!
bpalmer
rgr: Alice has 5 followers, including Bob. She says something clever or informative, which Bob thinks is interesting. He retweets it to his 500 followers.
s/thinks is interesting/thinks is worth propagating/
fledermaus
but what about Carol and Ted?
rgr
lets start from th ebeginning. What are those RT lines doing and how did they get generated and where are they going, if anywhere?
what emacs twitter page?
bpalmer
rgr: http://twitter.com/emacs
jordanb_
Then eve read Bob's twit stream and realized that he is totally vapid and his messages to Alice totally not worth unencrypting.
rgr
right and those lines are not there. Or were they just a joke?
bpalmer
rgr: yes.
jlf
obviously you're not a golfer
rgr
oh so there is no retwee bot
« prev 1 2 3 4 5 6 7 next »