bug#7761: The NS port is still 11 times slower than normal

bug#7761: The NS port is still 11 times slower than normal

[Leo]

Hi all,

There are numerous bug reports on the issues of NS port, in particular
the performance degradation after turning flyspell on.

I did a build of the NS port from trunk two days ago. And I tried
flyspell-buffer on the about page (C-h C-a). And it was about 11 times
slower than a same run on other terminals (X11, tty, Mac) etc.

Some of the reports are as old as two years ago. There seems little hope
it will shape up in the foreseeable future. I have been told even as it
stands today it is already much improved.

Meanwhile, we have a better alternative (the macport¹) that provides
comparable feature set as in GNU/Linux and without any compromise in
performance.

I wonder if macport should be merged?

Thanks.

Leo

Footnotes: 
¹  ftp://ftp.math.s.chiba-u.ac.jp/emacs

bug#7761: The NS port is still 11 times slower than normal

[Jan Djärv]

A way to reproduce starting from -Q would be more helpful if this is to be 
fixed.  Enumerating the missing future set in the current port is also 
helpful, sweeping statements is not.

I don't think switching is an option, the macport doesn't seem to be based on 
emacs 24 yet.

	Jan D.


Leo skrev 2010-12-30 11.05:
> Hi all,
>
> There are numerous bug reports on the issues of NS port, in particular
> the performance degradation after turning flyspell on.
>
> I did a build of the NS port from trunk two days ago. And I tried
> flyspell-buffer on the about page (C-h C-a). And it was about 11 times
> slower than a same run on other terminals (X11, tty, Mac) etc.
>
> Some of the reports are as old as two years ago. There seems little hope
> it will shape up in the foreseeable future. I have been told even as it
> stands today it is already much improved.
>
> Meanwhile, we have a better alternative (the macport¹) that provides
> comparable feature set as in GNU/Linux and without any compromise in
> performance.
>
> I wonder if macport should be merged?
>
> Thanks.
>
> Leo
>
> Footnotes:
> ¹  ftp://ftp.math.s.chiba-u.ac.jp/emacs
>
>
>

bug#7761: The NS port is still 11 times slower than normal

[Leo]

Sorry I was going to send this to the devel list.

On 2010-12-30 11:36 +0000, Jan Djärv wrote:
> A way to reproduce starting from -Q would be more helpful if this is
> to be fixed.  Enumerating the missing future set in the current port
> is also helpful, sweeping statements is not.

Could you reproduce it by:

 1. C-h C-a
 2. M-x flyspell-buffer

I could see the point moving slowly from word to word. A similar run on
tty or X11 would be nearly instant.

Sorry I cannot try it again since I have already deleted the build. It
associates tons of files with Emacs and deleting is the only way I know
that could quickly remove the associations.

I haven't been using the NS port for over 15 months. I can't remember
what is missing. But I remember it gave me enough reasons to build a
patched emacs from source for the very first time on OSX.

But the author of macport has detailed its features here:
http://permalink.gmane.org/gmane.emacs.devel/124489

Hope this useful.

> I don't think switching is an option, the macport doesn't seem to be
> based on emacs 24 yet.

I think merging it isn't too difficult.

Leo

bug#7761: The NS port is still 11 times slower than normal

[Glenn Morris]

I don't see a need to keep open this particular report, which was marked
"wontfix" some time ago.

This is a very general report. It is better to focus on specific issues
in separate reports and try and fix them (for example, #2056).

bug#7761: The NS port is still 11 times slower than normal

[Chong Yidong]

Leo <sdl.web@gmail.com> writes:

> Some of the reports are as old as two years ago. There seems little hope
> it will shape up in the foreseeable future. I have been told even as it
> stands today it is already much improved.
>
> I wonder if macport should be merged?

Or, use the same time and effort to fix the bugs in the NS port.