Re: [reSIProcate] resip footprint

[Alan Hawrylyshen <alan@xxxxxxxxxx>]

On Jul 22, 2004, at 08:19, martin.van.den.berg@xxxxxxxxxxx wrote:

> Hello all,
>
> From the mail list archive i've found this message conserning resip' 
> footprint.
>
> http://list.sipfoundry.org/archive/resiprocate-devel/msg00928.html
>
> The total, accoring to this writing is approx 2,5 Mb
>
> However, if I compile resip using VS7, the resiprocate.lib results in 
> approx 12Mb. I'm flabbergasted... these numbers are lightyears apart. 
> Can anybody explain me the difference?
>
> Thanks!

Martin;

The quoted number is for a resiprocate library built using gcc 3.2.2 
under GNU/Linux for the i686 architecture.
The number is also the size of the library text segment, (or roughly, 
the size of the _stripped_ library).
Due to the long, verbose nature of C++ mangled symbols, the unstripped 
library is routinely > 10Mb.
Rest assured that compiled for a target, and stripped, the library 
shrinks by an order of magnitude.

Here's a quick pre-0.5.0 snapshot from one of my views:
alan@tupolev .libs $ for N in *stripped; do bn=$(basename $N 
.stripped); ls -ls $N $bn; done
33072 -rw-r--r--    1 alan     eng      33821016 Jul 16 23:45 
libresiprocate.a
 2000 -rw-r--r--    1 alan     eng       2041902 Jul 22 08:43 
libresiprocate.a.stripped
15824 -rwxr-xr-x    1 alan     eng      16179360 Jul 16 23:45 
libresiprocate.so.0.0.5
 1948 -rwxr-xr-x    1 alan     eng       1989276 Jul 22 08:43 
libresiprocate.so.0.0.5.stripped

I realise that this isn't directly the footprint, but it goes a long 
way towards explaining the disconnect between library file size and 
resident code size.

Cheers,

Alan

Alan Hawrylyshen
Jasomi Networks Inc.
http://jasomi.com/
a l a n a t j a s o m i d o t c o m