Re: [PATCH 1/4] stringbuf: A string buffer implementation

[Rusty Russell <rusty@rustcorp.com.au>]

On Saturday 27 October 2007 06:57:14 Matt Mackall wrote:
> Well I expect once you start letting people easily build strings by
> concatenation, you'll very shortly afterwards have people using them
> in loops.  And having hidden O(n^2) behavior in there is a little sad, 
> even though n will tend to be small and well-bounded. If we can do
> something simple to avoid it, we should.

Hi Matt,

        I avoid typing even a single character of optimization until it's 
justified.  This is partially a reaction against the machoptimization 
tendencies of many kernel programmers, but it's mainly a concern at the 
kernel's complexity creep.

Meanwhile, of course, I've now spent far too long analyzing this :)

Building a 1000 byte string 1 byte at a time involves 6 reallocs (SLAB) or 10 
reallocs (SLUB).  Frankly, that's good enough without an explicit alloc 
length field (better in some ways).

As to keeping an explicit length vs strlen(): those 1000 calls on my test 
machine take 1491ns per call with an explicit length vs 1496ns per call with 
strlen().  That's not worth 4 bytes, let alone a single line of code, O(n^2) 
or no.

As the nail in the coffin, callers only use ->buf, so are insulated from any 
such optimizations if we decided to do them in future.

Hope that helps,
Rusty.
PS.  I don't think we should switch this to a simple char ** tho, as 
the "struct stringbuf" gives us some type safety and reminds people not to 
simply kfree it.