Caolán McNamara wrote:

> On Thu, 2008-09-25 at 23:41 +0200, Mathias Bauer wrote:
>> Caolán McNamara wrote:
>> 
>> > On Thu, 2008-09-25 at 15:11 +0200, Mathias Bauer wrote:
>> >> But the "Exit" method now is never called and IMHO it isn't necessary to
>> >> do so. The only known problem this creates is that it prevents the
>> >> detection of memory leaks caused by the objects that now never get
>> >> deleted. 
>> > 
>> > Possibly still the theoretical option of hooking them back into the
>> > library unloading scheme of
>> > http://udk.openoffice.org/common/man/spec/library_unloading.html
>> 
>> IIRC this was not meant to be applied to libraries that inject pointers
>> to objects implemented inside them into other libraries' code, they
>> where meant to be used for libraries where the life cycle of each object
>> can be tracked by UNO reference counting.
> 
> Well, I only poked around starmath (as I reckoned it would be the
> simplest user) where the only usage is inside SmDocument_createInstance
> which is the uno entry point. So that would make it a very good
> candidate, in that the only way to get trigger the Init is when
> creating a uno referenced object, so if all the starmath components 
> ref/def-ed the module-used count and if something existed to call the 
> library unloading-foo then it might in theory work out.

All SFX based modules implement an object deriving from SfxModule that
is inserted into an array inside of SFX. It's possible that other
objects also are inserted into such arrays. So before we could think
about using the Exit() methods we had to make sure that all of these
gaps have been removed.

This indeed would be a nice thing - it would improve our architecture a
lot.

Regards,
Mathias

-- 
Mathias Bauer (mba) - Project Lead OpenOffice.org Writer
OpenOffice.org Engineering at Sun: http://blogs.sun.com/GullFOSS
Please don't reply to "[EMAIL PROTECTED]".
I use it for the OOo lists and only rarely read other mails sent to it.

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to