httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brian Havard" <>
Subject Re: [PATCH] buffered reading of the cfg file
Date Wed, 14 Jun 2000 17:06:54 GMT
On Wed, 14 Jun 2000 12:46:54 -0400, Jeff Trawick wrote:

>> From:
>> Date: Wed, 14 Jun 2000 08:47:27 -0700 (PDT)
>> > The OS/2 code works fine and is the original version. I did the Win32 code
>> > too but commited it with the comment that it was only tested to compile,
>> > not work. As this was some time ago & nobody's complained I'd assumed it
>> > works. If it doesn't let me know, it should only require minor tweaks. I
>> > think Ryan did the unix version.
>> I did do the Unix version, it it too was commited with the comment that it
>> was only tested to compile and not work.  I haven't really looked at this
>> code again.  I find it hard to believe that this code requires the major
>> overhaul that was posted, because basically the code is working on
>> OS/2.
>> Ryan
>Which part of the change is too much?

Err, the fact that there's so much of it? :)

>At the expense of a little speed, ap_fgets() could call ap_read()
>instead of walking through the buffer itself, but if we can justify
>ignoring '\r' on Unix I'd rather ap_fgets() look for '\n' with
>memchr() anyway so that it can jump through the buffer (as memchr()
>and memcpy() tend to turn into inline code with compiled with

Interesting idea, though it has it's own complications like handling when
the buffer doesn't have the whole line in it. Still, it would be useful to
benchmark both methods & see if there's significant speed gain to be had.

 |  Brian Havard                 |  "He is not the messiah!                   |
 |  |  He's a very naughty boy!" - Life of Brian |

View raw message