Re: [Gnash-dev] Memory allocation model?

[Michael Meier]

> [gnash-dev put back for general interest]

Thanks! Wrong address...

> 
> On Tue, Nov 21, 2006 at 10:44:49PM +0100, Michael Meier wrote:
> 
> 
>>>libtool --mode=execute valgrind ../../gui/gnash -1 -r0 -v String.swf 2> ERR
>>
>>Thanks, did this.
> 
> 
> we might still have problems with the above, see 
> https://savannah.gnu.org/bugs/?18341
> 
> 
>>This is the output with String.swf checking and though calling my new
>>methods.
>>==5502== LEAK SUMMARY:
>>==5502==    still reachable: 103180 bytes in 151 blocks.
> 
> 
>>And this is without:
>>==5451== LEAK SUMMARY:
>>==5451==    still reachable: 94884 bytes in 151 blocks.
> 
> ..
> 
>>So I guess it does nasty things? :/
> 
> 
> Not necessarly, as the allocated memory is *still reachable*
> (there are pointers to them)
> 
> First of all run with GLIBCPP_FORCE_NEW=1 set in the environment.
> This will make sure libstdc++ (assuming you're using the GNU version)
> won't keep memory allocated on delete.
> 
> Second, you implemented something new, so you're likely returning
> String values in the environment, so I'd expect more memory to
> be used there (more ActionScript variables will have real values rather
> then being undefined).
> 
> As long as no *suppressed* errors reported you're doing nothing nasty.

Ah, well, it's
ERROR SUMMARY: 6 errors from 6 contexts (suppressed: 135 from 1)
in both cases.

Thanks much for your explanations! Hehe, I learned a lot today :D
I'll have a look at it again as soon as possible!


cheers,
Michael