Re: [Matlab-emacs-discuss] matlab-emacs slows down

[Eric L. <Eri...@ma...>]

Hi,

I don't recall why we picked 8192 as the size of the back-search buffer.   If some other value makes sense, I'll be happy to update the code with the revised smaller number.

Of course, with the newer MATLAB I tried out today, it no longer emits the hyperlink text at all, so it is searching for stuff it will never find.   As such, just removing the feature is probably for the best.   I had to add it in the past because the buffer used to be full of messy text.  In its place, we will need to identify the new output pattern, but we won't need to do preemptive hyperlinking to do it.

Thanks
Eric

From: Odd Andersen [mailto:odd...@gm...]
Sent: Thursday, March 06, 2014 6:38 AM
To: Sina Tootoonian
Cc: mat...@li...
Subject: Re: [Matlab-emacs-discuss] matlab-emacs slows down

Hi Sina,
That does sound like the function I was talking about.  In my matlab.el version (2012), there was no harcoded limit, so the problem with this function was perhaps more severe than in your case.  If you still have performance issues, the elisp profiler is a useful tool to pinpoint the problem.
Odd A.

On Thu, Mar 6, 2014 at 12:22 PM, Sina Tootoonian <sin...@gm...<mailto:sin...@gm...>> wrote:
Dear Odd and Ondra,
I did manage to find an earlier (2011) discussion (propmpted by Ondra) of a very similar problem, with very similar functions being implicated (rendering error markup). The specific problem was that a re-search-backward call was being made with a nil second argument, forcing it search the buffer all the way back. The fix at the time was to hard-limit the search. In the ~2013 version of matlab.el that I have, the search was limited to 8192 chars back. As a temporary  measure I've limited it even further to ~800 chars, to see if it makes a difference. Will let you know if it does, and if not, will try your solution Odd.

Thanks for your help and suggestions,

Sina

2014-03-06 9:22 GMT+00:00 Odd Andersen <odd...@gm...<mailto:odd...@gm...>>:

Hi,

I used to have the same problem, and found it frustrating that there seemed to be no discussion of this issue anywhere.  Finally, I was able to fix it myself.

Here's what I did:

  1.   I ran matlab-shell with my usual workflow until it started becoming unresponsive.
  2.  I started the emacs profiler: (profiler-start 'cpu)
  3.  I hit enter a couple of times in the matlab-shell, waiting a few seconds for the shell to respond.
  4.  I asked the profiler for a report (profiler-report)
  5.  From the report, it turned out that most of the runtime was spent in a single function whose task is to render errors as anchors (which I don't particularly care about).
  6.  This function is hooked into matlab-shell-mode when it start.  As a workaround, I added another hook that removes this hook, by including the following lines in my .emacs file:
(add-hook 'matlab-shell-mode-hook
      (lambda () (remove-hook 'comint-output-filter-functions
                  'matlab-shell-render-errors-as-anchor t)))
I have not had any issues with unresponsive matlab shell after I did this.  If your problem has the same cause as mine, this hopefully works for you too.  (Make sure to get the quotation marks in the above command right, as well as the final 't' on the last line...).

Odd A.


________________________________
-------- Original Message --------
Subject:

[Matlab-emacs-discuss] matlab-emacs slows down

Date:

Wed, 5 Mar 2014 18:51:01 +0000

From:

Sina Tootoonian <sin...@gm...><https://webmail.sintef.no/owa/redir.aspx?C=7u2jHFM45kiDABWqV6fI02wyMtT6B9EIn6FZnc1WYUOX0Gl19XWr3b0pnDSeinQ8CMAbLPXIWeM.&URL=mailto%3asina.tootoonian%40gmail.com>

To:

mat...@li...<https://webmail.sintef.no/owa/redir.aspx?C=7u2jHFM45kiDABWqV6fI02wyMtT6B9EIn6FZnc1WYUOX0Gl19XWr3b0pnDSeinQ8CMAbLPXIWeM.&URL=mailto%3amatlab-emacs-discuss%40lists.sourceforge.net>



Hi folks,



First, great job on the update to matlab-mode, a big improvement on

the version I was using before (~2003). One problem I've noticed is

that the CLI slows down considerably after a while, perhaps due to the

buffer getting large ( O(10k) lines). I usually end up restarting

matlab-shell to get the responsiveness back. Is there a better

workaround for this?



Thanks!



Sina



------------------------------------------------------------------------------

Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.

With Perforce, you get hassle-free workflows. Merge that actually works.

Faster operations. Version large binaries.  Built-in WAN optimization and the

freedom to use Git, Perforce or both. Make the move to Perforce.

http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk<https://webmail.sintef.no/owa/redir.aspx?C=7u2jHFM45kiDABWqV6fI02wyMtT6B9EIn6FZnc1WYUOX0Gl19XWr3b0pnDSeinQ8CMAbLPXIWeM.&URL=http%3a%2f%2fpubads.g.doubleclick.net%2fgampad%2fclk%3fid%3d122218951%26iu%3d%2f4140%2fostg.clktrk>

_______________________________________________

Matlab-emacs-discuss mailing list

Mat...@li...<https://webmail.sintef.no/owa/redir.aspx?C=7u2jHFM45kiDABWqV6fI02wyMtT6B9EIn6FZnc1WYUOX0Gl19XWr3b0pnDSeinQ8CMAbLPXIWeM.&URL=mailto%3aMatlab-emacs-discuss%40lists.sourceforge.net>

https://lists.sourceforge.net/lists/listinfo/matlab-emacs-discuss<https://webmail.sintef.no/owa/redir.aspx?C=7u2jHFM45kiDABWqV6fI02wyMtT6B9EIn6FZnc1WYUOX0Gl19XWr3b0pnDSeinQ8CMAbLPXIWeM.&URL=https%3a%2f%2flists.sourceforge.net%2flists%2flistinfo%2fmatlab-emacs-discuss>


------------------------------------------------------------------------------
Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.
With Perforce, you get hassle-free workflows. Merge that actually works.
Faster operations. Version large binaries.  Built-in WAN optimization and the
freedom to use Git, Perforce or both. Make the move to Perforce.
http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
_______________________________________________
Matlab-emacs-discuss mailing list
Mat...@li...<mailto:Mat...@li...>
https://lists.sourceforge.net/lists/listinfo/matlab-emacs-discuss