Re: Squid development

From: Joe Cooper <joe@dont-contact.us>
Date: Tue, 12 Jun 2001 18:20:09 -0500

Henrik Nordstrom wrote:

> Michael Y. Popov wrote:
>
>
>>1. Squid uses MD5 algorithm quite frequently, so MD5 main (which took
>>most of CPU power) function may be rewritten in assembly for better
>>performance. I does it already, using Netwide Assembler (freeware).
>>So code becomes little smaller and about 45% faster than code,
>>produces by C compiler (Microsoft Visual C++ 6.0, since I use SquidNT
>>ported by Romeo Anghelache). Only one disadvantage exists: this code
>>is only for 32-bit x86 architecture only. The code is highly optimized
>>for P6 architecture.
>>If you find this work useful for Squid, I'll send you code (asm file)
>>with 'MD5Transform' function.
>>
>
> Squid-2.5 can be linked with OpenSSL which includes MD5 routines which
> is substantially faster than the code shipped with Squid (45% faster).
>
> Shipping assembly language versions with Squid is a bit messy,
> especially considering that there are other free packages available that
> Squid may use...

Any reason this MD5 couldn't be brought into Squid? OpenSSL is under an
Apache style license, right? I'm not sure of the legality of
relicensing Apache-Style License to GPL...

>>2. Second idea is about Squid's IP and FQDN caches. All Squid versions
>>(known to me) use this caches only during uptime, so Squid restarting
>>cause lost of both caches. I think it will be useful to write these
>>caches to files and reloads its on 'init' phase of Squid loading.
>>Modifications if quite little and all basic code will be unmodified.
>>Unfortunately, I implemented this idea only for Win32 platform for
>>now. But I think re-porting to standard file I/O functions will not be
>>difficult.
>>
>
> Personally I like that the IP and FQDN caches in Squid are ditched on a
> restart, but I recognise that other may have other opinions on this.
>
> Fellow Squid developers: What is your opinions?

I don't see a benefit to saving it. Restarts are rare (every few
months, right? ;-) and these caches should be pretty much considered
non-permanent due to the nature of their contents.
                                   --
                      Joe Cooper <joe@swelltech.com>
                  Affordable Web Caching Proxy Appliances
                         http://www.swelltech.com
Received on Tue Jun 12 2001 - 17:10:20 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:03 MST