Any advantages to recompile Emacs...

Any advantages to recompile Emacs...

[Francis Moreau]

Hello,

I'm always using Emacs with the '-nw' switch that is in console mode.

I'm wondering if I can recompile Emacs without the graphic support and
if there're any improvements (perhaps starting faster ?) in doing so.

Thanks
-- 
Francis

Re: Any advantages to recompile Emacs...

[Peter Dyballa]

Am 10.12.2008 um 08:51 schrieb Francis Moreau:

> I'm wondering if I can recompile Emacs without the graphic support and
> if there're any improvements (perhaps starting faster ?) in doing so.


It's easy to do this: configure with --without-x. The binary will be  
smaller and of course it will be able to load faster. (Although I  
think GNU Emacs with windowing support can load at the same speed, so  
the smaller file size can significantly save some millisecs. Each time.)

--
Greetings

   Pete

You can learn many things from children.  How much patience you have,  
for instance.
				– Franklin P. Jones

Re: Any advantages to recompile Emacs...

[Francis Moreau]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 10.12.2008 um 08:51 schrieb Francis Moreau:
>
>> I'm wondering if I can recompile Emacs without the graphic support and
>> if there're any improvements (perhaps starting faster ?) in doing so.
>
>
> It's easy to do this: configure with --without-x. The binary will be
> smaller and of course it will be able to load faster. (Although I
> think GNU Emacs with windowing support can load at the same speed, so
> the smaller file size can significantly save some millisecs. Each
> time.)

Do you mean that 'emacs -nw' with X support is slower to start than
emacs without X support ?

thanks
-- 
Francis

Re: Any advantages to recompile Emacs...

[Peter Dyballa]

Am 10.12.2008 um 20:19 schrieb Francis Moreau:

> Do you mean that 'emacs -nw' with X support is slower to start than
> emacs without X support ?


Yes, every time you should be able to determine a few msec difference  
in launch time. The size of the binaries will vary, the number of  
shared libraries used and already residing in memory will vary – all  
the nsec and µsec saved will add ... every time!

This is from Apple's GNU Emacs without windowing support in Mac OS X  
10.4.11:

	time /usr/bin/emacs-21.2.1 -nw --no-init-file --no-site-file -kill
	
	        Time spent in user mode   (CPU seconds) : 0.043s
	        Time spent in kernel mode (CPU seconds) : 0.030s
	        Total time                              : 0:00.11s
	        CPU utilisation (percentage)            : 63.6%

My other self-compiled Emacsen (21.3.50, 22.1.50, 22.2, 22.3,  
23.0.50, 23.0.60) show something like this:

	time /sw/bin/emacs-21.3.50 -nw --no-init-file --no-site-file -kill
	        Time spent in user mode   (CPU seconds) : 0.103s
	        Time spent in kernel mode (CPU seconds) : 0.094s
	        Total time                              : 0:00.30s
	        CPU utilisation (percentage)            : 63.3%
	
or that:
	
	time /sw/bin/emacs-22.3 -nw --no-init-file --no-site-file -kill
	
	        Time spent in user mode   (CPU seconds) : 0.175s
	        Time spent in kernel mode (CPU seconds) : 0.076s
	        Total time                              : 0:00.33s
	        CPU utilisation (percentage)            : 72.7%
	time /usr/local/bin/emacs-23.0.60 -nw --no-init-file --no-site-file - 
kill
	
	        Time spent in user mode   (CPU seconds) : 0.201s
	        Time spent in kernel mode (CPU seconds) : 0.134s
	        Total time                              : 0:00.50s
	        CPU utilisation (percentage)            : 66.0%

All measurements after a few invocations if each variant. Each  
variant with windowing capabilities seems to initialise and restore  
the terminal emulation (some flashing/wiping effect). Could be GNU  
Emacs 19.28 is much much faster – at least in my memory and on my DEC  
VTxxx terminal at work ...

--
Greetings

   Pete

Well done is better than well said.
				– Benjamin Franklin

Re: Any advantages to recompile Emacs...

[Francis Moreau]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 10.12.2008 um 20:19 schrieb Francis Moreau:
>
>> Do you mean that 'emacs -nw' with X support is slower to start than
>> emacs without X support ?
>
>
> Yes, every time you should be able to determine a few msec difference
> in launch time. The size of the binaries will vary, the number of
> shared libraries used and already residing in memory will vary – all
> the nsec and µsec saved will add ... every time!

[...]

Sorry for digging out this thread, but I finally giving a try to build
emacs my self using the source code from CVS.

And one thing interesting is the size of the binary:

   emacs 22.2  with X support:  ~  8.0Mo
   emacs 23 without X support:  ~ 14.5Mo

It's a huge difference, so now I'm wondering if I correctly built the
beast...

Any Idea ?

Thanks
-- 
Francis

Re: Any advantages to recompile Emacs...

[Peter Dyballa]

Am 22.01.2009 um 20:44 schrieb Francis Moreau:

>    emacs 22.2  with X support:  ~  8.0Mo
>    emacs 23 without X support:  ~ 14.5Mo
>
> It's a huge difference, so now I'm wondering if I correctly built the
> beast...
>
> Any Idea ?

Think of Unicode (8 bit -> 32 bit)! And if your hardware is 64 bit,  
then GNU Emacs could have been configured and compiled for this,  
using pointers that large. Did you compile in debug information? (And  
GTK also consumes lots of disk space, look at my GNU Emacs 21.3.50  
(also GCC 3.3)!)

On my Mac the Emacsen have:

  7,490,192 26 Dez 01:23 emacs-22.3
10,166,896 14 Jan 15:27 emacs-23.0.60
13,836,700 20 Jul  2008 emacs-21.3.50


--
Greetings

   Pete

Programming today is a race between software engineers striving to  
build bigger and better idiot-proof programs, and the Universe trying  
to produce bigger and better idiots. So far, the Universe is winning.
				– Rich Cook

Re: Any advantages to recompile Emacs...

[Eli Zaretskii]

> From: Peter Dyballa <Peter_Dyballa@Web.DE>
> Date: Thu, 22 Jan 2009 22:19:47 +0100
> Cc: help-gnu-emacs@gnu.org
> 
> 
> Am 22.01.2009 um 20:44 schrieb Francis Moreau:
> 
> >    emacs 22.2  with X support:  ~  8.0Mo
> >    emacs 23 without X support:  ~ 14.5Mo
> >
> > It's a huge difference, so now I'm wondering if I correctly built the
> > beast...
> >
> > Any Idea ?
> 
> Think of Unicode (8 bit -> 32 bit)!

Unicode per se has no effect on the size of the Emacs binary, since
the internal representation of characters used by Emacs 23 is a
multibyte representation, as it was in previous versions of Emacs.

Re: Any advantages to recompile Emacs...

[Francis Moreau]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Think of Unicode (8 bit -> 32 bit)! And if your hardware is 64 bit,
> then GNU Emacs could have been configured and compiled for this,
> using pointers that large. 

Does emacs embed many unicode strings ??

> Did you compile in debug information?

I haven't asked for when doing the './configure' thing so I wouldn't
expect to have such info...

[me checking]

Well, this 'configure' thing compiled emacs with the debug info
(damn!), so after striping emacs, the size looks more reasonable:
9.9M.

> 
> (And GTK also consumes lots of disk space, look at my GNU Emacs
> 21.3.50 (also GCC 3.3)!)

Well I have asked to compile emacs without X support, so I don't think
GTK is the culprit.

thanks
-- 
Francis

Re: Any advantages to recompile Emacs...

[Peter Dyballa]

Am 22.01.2009 um 22:56 schrieb Francis Moreau:

> Does emacs embed many unicode strings ??

I don't know! 'strings <path to GNU Emacs> | wc -c' times 4 ... ?

>> (And GTK also consumes lots of disk space, look at my GNU Emacs
>> 21.3.50 (also GCC 3.3)!)
>
> Well I have asked to compile emacs without X support, so I don't think
> GTK is the culprit.

Yes, I read so. I used this as an excuse for 21.3.50's size.

--
Greetings

   Pete

If you don't find it in the index, look very carefully through the  
entire catalogue.
	–  Sears, Roebuck, and Co., Consumer's Guide, 1897

Re: Any advantages to recompile Emacs...

[Ian Eure]

On Jan 22, 2009, at 1:56 PM, Francis Moreau wrote:

> Peter Dyballa <Peter_Dyballa@Web.DE> writes:
>
>> Think of Unicode (8 bit -> 32 bit)! And if your hardware is 64 bit,
>> then GNU Emacs could have been configured and compiled for this,
>> using pointers that large.
>
> Does emacs embed many unicode strings ??
>
Probably not, and UTF-8 doesn't take any more space unless you use non- 
ASCII characters. For plain text, it's the same size.


>> Did you compile in debug information?
>
> I haven't asked for when doing the './configure' thing so I wouldn't
> expect to have such info...
>
> [me checking]
>
> Well, this 'configure' thing compiled emacs with the debug info
> (damn!), so after striping emacs, the size looks more reasonable:
> 9.9M.
>
The default behavior of GCC is to include debugging symbols. You need  
to explicitly remove them if they're not needed.

  - Ian

Re: Any advantages to recompile Emacs...

[Francis Moreau]

On Fri, Jan 23, 2009 at 12:14 AM, Ian Eure <ian@digg.com> wrote:
>
> The default behavior of GCC is to include debugging symbols. You need to
> explicitly remove them if they're not needed.

I don't think so, you need to pass the famous '-g' switch for that
(fortuantely).

Thanks
-- 
Francis

Re: Any advantages to recompile Emacs...

[Alberto Luaces]

El Viernes 23 Enero 2009ES 09:26:22 Francis Moreau escribió:
> > The default behavior of GCC is to include debugging symbols. You need to
> > explicitly remove them if they're not needed.
>
> I don't think so, you need to pass the famous '-g' switch for that
> (fortuantely).

Anyway, you should have a "make install-strip" target available in the 
configure generated Makefile.

Alberto