Index: [thread] [date] [subject] [author]
  From: Stefan Mars <mars@lysator.liu.se>
  To  : ggi-develop@eskimo.com
  Date: Mon, 12 Apr 1999 12:40:10 +0200 (MET DST)

Resent from berlin: woo hoo (#2) (fwd)

The following post just appeared on the berlin mailing list, and it has a
complaint about our event system. I haven't had a look at it for quite
some time, but maybe someone who is up to speed can comment?

/Stefan

---------- Forwarded message ----------
Date: Mon, 12 Apr 1999 06:24:14 -0400
From: Graydon Hoare <graydon@pobox.com>
To: design@berlin-consortium.org
Subject: woo hoo (#2)
Resent-Date: 12 Apr 1999 10:24:17 -0000
Resent-From: berlin-design@berlin-consortium.org
Resent-cc: recipient list not shown: ;

you remember last july I got the first working images out of the
previous version of the berlin server? Well, after a week + weekend of
what can only be described as pure unmitigated debugging heck, I have
beaten the second generation server into submission and it's now
drawing for us like a good little munchkin. I'm going to chew some
people out now. if you haven't been involved in either GGI or Berlin
code-wise, you can ignore the rest of the message.

(1) whoever wrote GGI's event system: it's not threadsafe, still! I
can't inject new events into the queue using ggiEventSend without
corrupting the queue structure; since this is the only reasonable way
to have 1 thread tell an event-reader it wants to do some drawing, 
we're back to a polling loop. this "grinds my gizzard"

(2) stefan: the mixing of _ptr types, _var types, and implemetation
pointers has got to stop! The compiler thinks they're the same, but
they're not, so something looks ok, and then falls over at runtime
when you use a remote object reference.

(3) whoever is maintaining gdb: any idea when it'll work with threads
and I can debug properly using, like, something other than print
statements?

(4) stefan: the _this() and _boa() methods are completely meaningless
in an object which has not been registered with a BOA. they return
junk. If you are going to construct sub-objects inside a constructor,
you need to use some other facility like CORBA::BOA::getBOA(), or
delegate to a post-construction init() method, or something.

(5) stefan: amazing work, BTW. I don't want to sound like I'm all
criticism, because this code really does do what it's supposed to
despite the problems. You're a god. You've single-handedly ported
everything which is good and pure in fresco to berlin, and we shall
all dance joyfully and get extraordinarily drunk in celebration of
your fine work.

now then, if y'all don't mind, it being 6:30 am on a monday morning, I 
think I'm gonna commit the code, and go get some nappy time.
-graydon


--  
To UNSUBSCRIBE, email to design-request@berlin-consortium.org
with a subject of "unsubscribe". Trouble? Contact listmaster@berlin-consortium.org


Index: [thread] [date] [subject] [author]