[coreboot] [PATCH] cbfs, smaller api, more types

[Carl-Daniel Hailfinger]

On 27.02.2010 16:09, Stefan Reinauer wrote:
> On 2/27/10 3:51 PM, Peter Stuge wrote:
>   
>> Stefan Reinauer wrote:
>>   
>>     
>>> Since we only do name based matching in coreboot anyways, do you
>>> suggest we drop the type field?
>>>     
>>>       
>> Well, yes, I think I am..
>>
>> I know there are cases when it's handy to inspect the type, but
>> unless the type is the _only_ thing that matters it isn't so
>> intuitive to have one at all.
>>
>> What do you think?
>>   
>>     
> * I think Kevin might not like that idea. He's using the type in SeaBIOS.
> * Maybe SeaBIOS can be changed? Who will do that?
> [...]
> So I think we should keep it for now and keep the possibility to drop it
> in mind.
>   

IMHO the time to change anything in CBFS is over. It is too widely used
to change the in-ROM format in a way that is not 100% backwards
compatible. Your patch might be backwards compatible, but some of the
proposed extensions (option ROM naming and separate PCI ID storage) are not.

OTOH, if we change the in-ROM format, we might as well fix the design
shortcomings I mentioned back in the LAR+SELF debate. AFAIK modern CBFS
still is a stripped down LAR+SELF.

My sincere apologies if I missed some important development or
misunderstood the proposed changes.

Regards,
Carl-Daniel

-- 
"I do consider assignment statements and pointer variables to be among
computer science's most valuable treasures."
-- Donald E. Knuth