[Scummvm-devel] ScummVM 0.12.0: Release status -- 2008-07-13

[Max Horn]

Am 14.07.2008 um 16:03 schrieb John Willis:

>> Ports status:
>>
>> GP2X: ?
>
> Port builds and is in a fair state to my eyes.
>
> Starting to suffer under the strain of all the engines but I guess  
> it's time
> for a split build as I can't seem to get the dynamic plugin setup to  
> play
> nice with a static executable (and am rather low on free time right  
> now).

Clearly, the binary size is an issue for more and more porters. We  
will keep adding engines, so I am afraid the only real way out will be  
for as many ports as possible to implement dynamically loadable  
plugins. Again, I'll be happy to assist with this where possible.

However, that shouldn't stop us from asking whether we can remove some  
slack in our engines. On my Intel MacBook Pro, with -O and without -g,  
I got the following "Top 5 Engines", in terms of their size:

#1: 2662916 scumm/libscumm.a (biggest by far, but then it supports  
lots of games)
#2: 1791444 kyra/libkyra.a   (clear 2nd place, almost twice as big as  
the 3rd one -- wonder why...)
#3:  922020 gob/libgob.a
#4:  785716 agos/libagos.a
#5:  703796 parallaction/libparallaction.a

And the top 10 *.o files of our engines:

143432 lure/hotspots.o
135416 kyra/staticres.o
127852 scumm/detection.o
124412 agi/detection.o
107680 scumm/he/script_v100he.o
102468 scumm/scumm.o
  88588 kyra/script_hof.o
  88396 touche/touche.o
  87604 kyra/script_lok.o
  85000 scumm/script_v6.o

I wonder what other compiles make out of this. Also, for the top four  
files, their size seems to a big degree to stem from lots of data  
tables they compile in.


Cheers,
Max