Re: [OpenBIOS] Defining ." and "

[Parag Patel]

In message , Segher Boessenkool writes:
> garbage collect.  ouch.
> in dictionary or similar --> never freed
> in "string buffers" -- either 1 or 2, but both problems
>
> less severe.

3.5) A fixed number of static fixed-length buffers that are treated as a
ring of least-recently-used buffers.  The Open Firmware spec requires at
least two.  We typically use two buffers for " and another two for s"
just to make sure that they don't step on each other.

The idea is that since only two are required by the spec, all
development code should be able to run with at least two, and thus
increasing the number later or for a specific port won't cause
portability problems (as going the other way would).

-- Parag
-
To unsubscribe: send mail to [e-mail]
with 'unsubscribe openbios' in the body of the message
http://www.freiburg.linux.de/OpenBIOS/ - free your system..

Defining ." and "

[Segher Boessenkool]

> 3.5) A fixed number of static fixed-length buffers that are treated as a
> ring of least-recently-used buffers.  The Open Firmware spec requires at
> least two.  We typically use two buffers for " and another two for s"
> just to make sure that they don't step on each other.
>
> The idea is that since only two are required by the spec, all
> development code should be able to run with at least two, and thus
> increasing the number later or for a specific port won't cause
> portability problems (as going the other way would).

Where does it say it requires two?  I can't find no reference to this in
the 1275 spec.  The ANS, on the other hand, requires only one, and even
then only *if* you implement S" in interpreting state.

Cheers,

Segher

-
To unsubscribe: send mail to [e-mail]
with 'unsubscribe openbios' in the body of the message
http://www.freiburg.linux.de/OpenBIOS/ - free your system..

Defining ." and "

[Stefan Reinauer]

* Segher Boessenkool  [020608 15:23]:
> Where does it say it requires two?  I can't find no reference to this in
> the 1275 spec.  The ANS, on the other hand, requires only one, and even
> then only *if* you implement S" in interpreting state.

Still, most (all that I know of) implementations can cope with S"/" in
interpreting state - might be better if we stay with that majority..

Stefan