[PATCH 0 of 1] context: reduce use of expensive cmp even further

[Nicolas Dumazet]

2010/10/15 Christian Ebert <blacktrash at gmx.net>:
> Hi,
>
> This might be overkill. On the other hand it checks for the
> presence of _encodefilterpats first, and for those who are
> encoding this might still be faster.

Well I don't know if we want this, but it's interesting to explore the
possibility.
I actually do not know so much about encode filters: just that I had
to avoid them for size checks because they are likely to change
content size.
Do we have a list/an idea of how/in which case they're supposed to be used?

If, in average, we have very few filters, then it might save us some
time. It would be worth to check what's happening.
Would you know typical use cases, Christian? Would you be able to run
some tests and come back with figures, to check if we're actually
saving time with this patch?

If we do save time, then of course I'd push this patch and its keyword brother.

Thanks!
>
> If we decide to go this way, I can implement the same in
> hgext.keyword.
>
> c
>



-- 
Nicolas Dumazet — NicDumZ