From: Brian Julin <bri@forcade.calyx.net>
  To  : ggi-develop@eskimo.com
  Date: Tue, 23 Feb 1999 15:01:35 -0500 (EST)

Re: Load-time parameters

On Tue, 23 Feb 1999, Marcus Sundberg wrote:

> Brian Julin wrote:
> Now for the reall comments:
> This is not good. It produces bloat both in size and cycles,

Most of the functions would be extremely simple knockoffs of a
template.  Passing of these parameters is NOT performace critical
and only happens during initialization stages.

> and is
> not well suited for hierarchical systems like Linux sysctl/proc or
> win32 registry.

How so?? just open a sysctl file and see the key,value pairs
or put the keys,values in a subfolder the registry.  The values are 
just prepacked C data.

> A somewhat better system is to have a structure which is filled in
> with parameters by the driver and passed to the KGI subsystem which
> fills in values.

That's a very vague description.  What do you mean?  I hope you're not
accusing me of wasting cycles and at the same time suggesting we have 
a parser in-kernel :).

> It might require som macro magic if we want to support
> systems like Linux insmod parameters where each parameter must be
> declarated at compile time, it's more flexible than a bunch of
> functions.

You're not making sense.  No macro magic will be more flexible
than a bunch of standard functions.  These are the lowest common 
denominator.  You have systems that will create symbols for you, per 
insmod, and systems that won't.  If you think you can patch over
this difference with "macro magic" then do so, but if it's so easy to 
do that way (I've actually tried) why's it been broken and unusable 
for almost a year now?

Suggest an alternative at least.

--
P.C.M.C.I.A. stands for "Plastic Connectors May Crack If Adjusted"
                                                             -- me
--
Brian S. Julin