On 12/17/06, Attila Kinali <[EMAIL PROTECTED]> wrote:

For the old stuff, i'd just create a new directory "old_stuff"
at the top level, then clone the current hierarchy we have and
"svn move" directories in there, best with a date attached. ie
cd <ogptree>/trunk
mkdir old_stuff
svn add old_stuff
svn mv pci old_stuff/pci-`date -I`
svn ci

Ok.  Sounds good.  For the moment, I'll just move the PCI stuff.  We
should adopt this policy for other things as we go along.  Never throw
things out, but always keep your working space nice and clean so
people can make sense of it.



> (3) Going with the spirit of being an educational project, not to
> mention good engineering practices, I think we should document this in
> fine detail.

Definitly. I once had a few hours free time and tried
to make sense of the current file structure and what
we have. I gave up after two hours of code reading.
Some READMEs that describe the contents of each directory
would really help

Yeah.  Somehow, we need to recruit some of the list members to take
charge of documentation.  They would spend their time asking the
developers questions, basically, and writing about what the developers
explained.  Anyone?  :)


> Next, we should discuss strategy.  I'm thinking that we urgently need
> a PCI target (the one we're testing with does not belong to us), so I
> think the next step is to make this synthesizable.  However, we should
> take a baby-step towards adding the master logic so that we don't have
> to do a lot of back-tracking when we do get around to that.

As i never thought about how to implement a PCI interface,
i don't have a clear idea what the difficulties are from
a target to a master. But couldn't we first stay with the
target, simulate it, synthesize it, test it fully and then
step by step add master functionalities?

The master is another state machine.  It's either there or not.
Actually, there are a few bits and pieces already there.  And then we
can add a non-functional place-holder to get the top-level resituated.
But then we need to suddenly do a lot of work to design the thing.
The thing that got me stuck was that I couldn't even draw a complete
state transition diagram for it.

Oh, BTW, here, I'm just talking about basic bus connectivity.  Things
like actually deciding what reads and writes are going to be done is a
seperate piece.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to