httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@kiwi.ICS.UCI.EDU>
Subject Re: [PATCH] Avoiding register long long
Date Wed, 01 Sep 1999 22:59:23 GMT
>Some compilers (noteably on 32 bit platforms) cannot deal with the whole
>functionality of the "long long" data types very well. One I have here
>cannot deal with the "register" option in conjunction with "long long"
>or "unsigned long long".
>
>Is there a very strong objection against the following patch, which just
>omits the "register" in combinations with "long long"? The source
>ap_snprintf.c has "register" attributes sprinkled all over, but the
>"long long" is used only very rarely in apache, so performance reasons
>are not all too important IMHO.
>
>What do you think? (Both 1.3 and 2.0)

+1 -- asking for the widest int is contrary to asking for a register, though
a valid compiler should just ignore the register modifier if it can't
be met. However, I don't know of any compilers that fail to optimize
something just because it doesn't have the register modifier, so there
is no harm in removing it IMO.

....Roy

Mime
View raw message