Re: [PATCH][merge-tracking]Better way to iterate the locally scoped hashes

Daniel Rall writes:
> On Mon, 22 Jan 2007, Peter Lundblad wrote:
>
So, about this apr_hash_first(pool versus NULL thing:

We have about 200 instances of apr_hash_first in the core code. All
use a pool except for svn_hash_clear, which has no pool available and
which obviously can't have other threads access the hash table anyway.
So we are pretty consistent in passing in a pool don't have to think about it.

> I doubt it buys us much in terms of memory footprint, but examing the
> flip side of the coin, what does it cost us? For local variables, I
> find the complexity cost to be negliable. (For non-local variables, I
> wouldn't favor this approach.) One could argue that this difference

I really think consistency, even regarding this minor issue, helps
maintainability. And soon we will start having those patches converting
local variable hash table iterations to use the internal hash table
iterator (which we already have, since this patch is such a one) just for
no visible gain at all.

Thanks,
//Peter

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jan 23 09:24:43 2007