[ESS] Tips for preventing slowing down?

[Marc Schwartz]

Hi,

To throw out one, potentially arcane possibility, in addition to Martin's comment below.

Check your .emacs file for the following entry:

  '(global-linum-mode t)

This is to display line numbers in the buffer frame, in addition to the mode line, if used.

With Emacs 26.x, this should not be used, rather you should use:

  '(global-display-line-numbers-mode t)

My recollection is that when I made the switch to Emacs 26 last year, I still had the former entry in my .emacs file.

I had noted the progression of a slowdown with large buffers, both in terms of general scrolling, as well as during the Sweave related processing of a large .Rnw file.

The latter entry is part of Emacs 26.x, whereas the former goes back to circa Emacs 22/23, I believe. The position of the line numbers in the buffer frame is also different now, than with the prior linum mode.

Through a lot of trial and error, I found this to be the etiology of my slow scrolling behavior and I am back to a speedy Emacs, even with files of 10s of thousands of lines. There was some conflict between the older linum mode and Emacs 26.x.

Regards,

Marc Schwartz


> On Nov 13, 2019, at 3:14 PM, Martin Maechler via ESS-help <ess-help using r-project.org> wrote:
> 
> Thanks a lot Bill.
> 
> I've observed the same... and so have others,  often very experienced R users.
> 
> One thing some of us found was that  'flymake' was very harmful in
> these situations,  so we have turned it off (or tried to ...)
> completely.
> But I have to acknowledge that it does not solve the problem in some cases:
> https://github.com/emacs-ess/ESS/issues/490#issuecomment-538682487
> -->    (setq ess-use-flymake nil)
> 
> Relatedly maybe (or not), I tend to have a suspicion that the "newish"
> ESS behavior of  "taking over" your default directory, and adapting
> behavior of  emacs *shell*   is also harmful, notably when that
> *shell* buffer also becomes 10000s of lines long.
> 
> On Wed, Nov 13, 2019 at 5:28 PM William Denton via ESS-help
> <ess-help using r-project.org> wrote:
>> 
>> I'm finding that large ESS buffers, where I've been running a lot of commands,
>> really slow down Emacs.  I use R a lot in Org, so when I'm working on something
>> I'm running little R code blocks over and over, and then regenerating the entire
>> document by running all the commands.
>> 
>> After a while, the ESS buffer gets large, and things get slow.  I mean really
>> slow, to where keystrokes get missed because of the lag.  I thought it might be
>> swiper and ivy, so I disabled them, but the problem still happens.  Running R
>> at the command line doesn't lead to slowdown.
>> 
>> I ran the profiler while I spent fifteen minutes or so in Org and ESS running R
>> commands, and the highlights are this:
>> 
>> - timer-event-handler                                           15189  34%
>>    cancel-timer-internal                                         8411  19%
>> 
>> - command-execute                                               14995  34%
>>  - call-interactively                                           14995  34%
>>   - funcall-interactively                                        8411  19%
>>    + org-export-dispatch                                         3822   8%
>>    + org-ctrl-c-ctrl-c                                           3702   8%
>>   - byte-code                                                    6583  15%
>>    + read-extended-command                                       6583  15%
>> 
>> - ...                                                           12560  28%
>>    Automatic GC                                                 12552  28%
>> 
>> I don't know what to make of this, though, or if there's anything there that
>> might indicate what I should do.  Do others have the same experience that large
>> ESS buffers slow down Emacs?  Is there any advice about what might help stop
>> that?
>> 
>> Thanks,
>> 
>> Bill