Hi,

It's complex, and it's consistent with the style of the cache manager to be 
parsimonious with stack.  I'd have to dedicate more time than I can today to 
speak more clearly to elegance, but parsimony contributes to elegance, too.

Matt

----- Original Message -----
From: "Simon Wilkinson" <[email protected]>
To: "Matt W.Benjamin" <[email protected]>
Cc: "Felix Frank" <[email protected]>, "OpenAFS Devel" 
<[email protected]>
Sent: Thursday, July 2, 2009 12:12:26 PM GMT -05:00 US/Canada Eastern
Subject: Re: [OpenAFS-devel] Patch to prepare the client for addition protocols


On 2 Jul 2009, at 15:59, Matt W. Benjamin wrote:

> Hi Felix,
>
> Good point.

The question of stack frames is actually really complex. Whether it's  
worth adding additional calling depth (Felix's proposal adds a single  
frame, my proposed changes would add one more) in order to simplify  
readability and maintainability really depends on a number of factors.  
Without knowing how deep the stack is at the point that Fetch/Store  
Proc is called, how many variables the callers have allocated on the  
stack, and how tight we are for headroom already, we're not going to  
know whether this is safe or not.

But I still maintain, from a pure elegance perspective, that  
abstraction layers that meander across multiple concepts are untidy.

S.

_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel
_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel

Reply via email to