[PATCH] Remove nTableSize from HashTable? View as Markdown

Hello Matt,

if there is no advantage in doing so, why offer a patch?

Ok, you probably meant disadvantage. And the advantge of doing so is
probably a lower memory foot print. Did you by any chance run any
profiling? Without numbers, I guess it is not worth the effort changing
a core component.

Sunday, March 15, 2009, 5:22:06 PM, you wrote:

Hi Dmitry, all,

Just a quick one... ;^) Is there much, if any, advantage to removing the
HashTable member nTableSize? It's just nTableMask + 1, so it can be derived
easily in the couple places that need it... Figured it can't hurt to make
things smaller, and it only adds a couple addition operations for resizing,
basically. :-)

Patches if you want the change (or tell me to commit):
http://realplain.com/php/tablesize.diff
http://realplain.com/php/tablesize_5_3.diff

• Matt

Best regards,
Marcus

Hi Matt,

I don't see a lot of reason in this patch as the benefit is near
invisible and readability becomes worse.

Also the patch can break third-party modules like APC and Zend
extensions. Even it's not a big problem as they can be easily fixed, I'm
against of changing of core components without real benefit.

Thanks. Dmitry.

Matt Wilmas wrote:

Hi Dmitry, all,

Just a quick one... ;^) Is there much, if any, advantage to removing
the HashTable member nTableSize? It's just nTableMask + 1, so it can be
derived easily in the couple places that need it... Figured it can't
hurt to make things smaller, and it only adds a couple addition
operations for resizing, basically. :-)

Patches if you want the change (or tell me to commit):
http://realplain.com/php/tablesize.diff
http://realplain.com/php/tablesize_5_3.diff

• Matt