FreeLists / openbeos / [openbeos] Re: A new BMessage implementation (Message4)

["Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>]

mmlr@xxxxxxxx wrote:
> > Now that the message reader is supposed to work, I should
> > find out why  the Deskbar still doesn't show the logo (or
> > any other bitmaps) when compiled under Dano - also with the
> > new build system.
> Does the Deskbar really store it's logo as a BMessage? I thought it 
> was
> a bitmap resource.

It's a bitmap resource - but a bitmap resource is a flattened BBitmap, 
and that is just a BMessage.

> Actually the reader is only supposed to work ;). As I stated it is 
> not
> tested.
> Is it possible that this:
> 
>     int32 count = *(int32 *)fieldBuffer + dataOffset;
> 
> is actually wrong and should be more like that:
> 
>     int32 count = *(int32 *)(fieldBuffer + dataOffset);

At least that doesn't change anything, even though it looks more 
correct. In any case, I commit this version.

> It just occured to me as I read my commit over. I cannot test that 
> right
> now as I'm not at home.

I just tried "mar" from Haiku ("mar -rr /system/Deskbar"), and the 
output looks different from running the same command on Dano (on the 
same executable). The Dano message reader obviously doesn't work 
perfectly yet (but probably a lot better than my previous version).

> > Do we want to make it the default already?
> It depends; if you want to take another risk right now. Currently we
> have bugs comming in because of the new build system, the app_server
> refactoring and the new clipping code. It gets kind of difficult to
> tell what is what. Especially since a message that is not delivered 
> can
> very well look like something in the app_server doesn't work. I'd 
> rather
> test and debug it some more. There is at least one known issue with
> Message4 and I havn't havily tested the R5 reader/writer just yet.

You're right, it makes sense to postpone the change a bit.

Bye,
   Axel.