W3C home > Mailing lists > Public > public-web-perf@w3.org > September 2010

Re: Deflate is Superior to Gzip (edited post)

From: David Murdoch <david@vervestudios.co>
Date: Thu, 16 Sep 2010 14:06:36 -0400
Message-ID: <AANLkTinJOU_8OaWobBurFEu9BauP9AbovSAeYxr1iJwg@mail.gmail.com>
To: Bryan McQuade <bmcquade@google.com>
Cc: public-web-perf@w3.org
I've read that proxies may have some trouble with compressed content
(don't they always) but haven't seen any real data. I have no idea how
to even begin testing for proxies, nor do I get enough web-traffic at
the moment to get an accurate snapshot of the proxy landscape anyway.

Any ideas?

David Murdoch | Verve Studios, Co
ph: (407) 374-3003 | fx: (407) 696-3078 | e: david@vervestudios.co |
web: http://vervestudios.co



On Thu, Sep 16, 2010 at 12:53 PM, David Murdoch <david@vervestudios.co> wrote:
> Yes, it is a wrapper around raw deflate. But that wrapper has a
> checksum that has to be calculated which also takes time on the
> server. If developers are taking time to optimize their CSS selector
> chains they will probably take time for an optimization such as using
> raw deflate over gzip.
>
> Personally, I don't think it is risky to switch. But that is what the
> compression test is trying to figure out. So far, RAW deflate IS
> supported in all browsers and platforms tested when "deflate" is
> present in the Content-Encoding request header.
>
> Do you have any data on RAW deflate failing?
>
> David Murdoch | Verve Studios, Co
> ph: (407) 374-3003 | fx: (407) 696-3078 | e: david@vervestudios.co |
> web: http://vervestudios.co
>
>
>
> On Thu, Sep 16, 2010 at 12:41 PM, Bryan McQuade <bmcquade@google.com> wrote:
>> Isn't gzip just a wrapper around raw deflate? So using raw deflate
>> will save you a small fixed quantity of bytes (~25 bytes IIRC)?
>>
>> I'm with you that if we were designing HTTP from scratch it would make
>> sense to pick raw deflate instead of gzip, but given that gzip is
>> widely supported and used, and switching is potentially risky and
>> saves only a handful of bytes per response, I'm not sure it's worth
>> changing at this point.
>>
>> On Thu, Sep 16, 2010 at 10:30 AM, David Murdoch <david@vervestudios.co> wrote:
>>> Sorry for the double post, the hyperlink in the first post was
>>> pointing to the dev version on my machine. The hyperlink has been
>>> corrected and now points to the production version of the test.
>>> ---
>>> Some very influential sources have been promoting the gzip compression
>>> format as the end-all and be-all to our HTTP 1.1 compression needs;
>>> some tout gzip as the superior compression format ("Gzip is the most
>>> [...] effective compression method..." [source: Best Practices for
>>> Speeding Up Your Website]). This, however, is not necessarily true.
>>> There are 2 other compression formats commonly available for use on
>>> the web.
>>>
>>> Research is currently being conducted at
>>> http://www.vervestudios.co/projects/compression-tests/results.
>>>
>>> Feedback and comments are encouraged.
>>>
>>>
>>
>
Received on Thursday, 16 September 2010 18:07:40 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:04:29 UTC