[interfacekit] Back in the saddle!

Hey, all.  My apologies for the lengthy absence -- I've been in mail 
*and* job hell the last few weeks. =P  Anyway, I seem to be back in 
operation, mail-wise.

Right out of the gate:  a big congratulations to DarkWyrm on Candace's 
birth!  Is she your first child?

Second up:  very tantalizing screenshot there, Cedric.  Makes me very 
antsy for the day our app_server can do that! =)

DarkWyrm, I don't know if you've dealt with the observer/notification 
issue, but I have some ideas that might be useful if you haven't, so do 
let me know.

On BBlockCache:  my suspicion is that the original implementation will 
only allocate up to a certain amount of memory, and just silently fails 
above that (without bothering to be clean about it, apparently).  I 
figure you could wire up a test that makes it allocate progressively 
larger chunks of memory until you get your segment violation -- and 
there's your top end.  Then the question becomes this:  do we want to 
match BBlockCache there, or make ours go all the way?  Whichever we do, 
a Get() on a failed allocation attempt should return NULL.  Also, with 
regards to unchecked_malloc, IIRC, this is simply a "regular" malloc, 
i.e., one that doesn't plug into the malloc_debug facility.  
Unsurprisingly, checked_malloc does. ;)  Generally speaking, my policy 
is that fixing something which currently fails 
ungracefully/catastrophically (e.g., seg fault) so that it fails nicely 
is a *good* thing.  Little stuff like that will make OBOS R1 more stable 
than R5, which is a bit of extra value-added incentive for users.

On BList & BListView:  As somebody mentioned there is the class in 
ObjectList which is templatized.  Don't bother trying to do this with 
BListView & Co. at this time.  Just stick with void* (as ugly as it is) 
and we'll revisit the issue for R2.  The sticky part is that the list 
view classes would probably also have to be templatized for it to be 
useful and that's a nasty can o' worms.  As far as BListView's handling 
of NULL arguments to various functions, what exactly are you seeing?  Is 
it a crash problem or just a silent failure?  If it's a silent failure, 
what return code is coming back?  I'm pretty definite on fixing it if 
it's a crasher -- it certainly shouldn't damage anybody's code to do 
that. ;P  As far as the horizontal scrollbar issue is concerned, it may 
be that folks have worked around the problem, so asking around is a good 
idea.

Issac, I'm a bit confused as to the BList::operator=() problem; it 
sounds like you're saying that the pointed-to items are copied here, not 
just the pointers themselves -- is that right?  The BeBook pretty 
clearly states that just the pointers are copied, but maybe that's not 
what you're seeing in testing ...  Wish I could offer something useful 
re. BRegion cleanup, but as a psychology major, math was *not* my strong 
suit. ;)

As for task ownership, go to the team pages and look at the schedule -- 
those items which have been assigned will have an "Owner" listed in the 
right column.  If you're working on something and I haven't got you as 
the owner (or some imposter listed instead ;), please let me know so I 
can update!  And, of course, let me know of any other problems that may 
crop up.

What have I been up to?  Well, aside from being whipped like a vagrant 
dog by my current work project, I've been putting in some hours on a 
tool to generate passthru wrapper classes.  The idea is this:  grind out 
the public interface of libbe.so which we haven't implemented yet.  Wrap 
this and our stuff up in an .so.  Create a symlink which test apps 
"link" to; this symlink points alternately to our lib or libbe.so.  This 
should help us quite a bit in tracking down places where we're not being 
binary compatible.  The work whipping should be over at the end of this 
week, and I'll have more time to put in after that.

Man, it's good to have mail again!

e

Our chief want is someone who will inspire us to be what we know we 
could be.
        - Ralph Waldo Emerson, writer and philosopher (1803-1882)

Other related posts: