[OpenAFS-devel] Patch to prepare the client for addition protocols

[Simon Wilkinson]

On 2 Jul 2009, at 17:28, Matt W. Benjamin wrote:

> Hi,
>
> It's complex, and it's consistent with the style of the cache  
> manager to be parsimonious with stack.

Whilst the cache manager may have historically been comprised of a  
small number of monolithic functions, I don't think it's really true  
to say that its stack usage has been sparing - just look at the amount  
of stuff it allocates on the stack. In any case, recent changes are  
already moving away from that style - as repeated code chunks  
gradually get refactored into single, maintainable segments, and new  
uses split existing functions into smaller chunks.

I think we should be cautious about radical increases in depth, but I  
don't think we should be constrained purely by consistence with the  
historical artifact that is the current cache manager's style. If we  
were, we'd still have everything contained in about 4 files ...

S.