W3C home > Mailing lists > Public > www-style@w3.org > March 2015

Re: "fast vs complete" is "SAX vs DOM"? and the IDs?

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Mon, 9 Mar 2015 14:39:01 -0700
Message-ID: <CAAWBYDDo-vDwm2SV9paRvXu6ypK+Rh4QNDkGOVt52dgixXvV4Q@mail.gmail.com>
To: Marat Tanalin <mtanalin@yandex.ru>
Cc: Boris Zbarsky <bzbarsky@mit.edu>, "www-style@w3.org" <www-style@w3.org>
On Mon, Mar 9, 2015 at 2:24 PM, Marat Tanalin <mtanalin@yandex.ru> wrote:
> 09.03.2015, 04:49, "Boris Zbarsky" <bzbarsky@mit.edu>:
>> On 3/8/15 9:22 PM, Marat Tanalin wrote:
>>>  Could you provide an estimation of how much faster should computers get compared with current-generation computers to drop "fast"/"complete"-profile separation?
>> It depends on the size of pages.  The issue with fast vs complete is
>> that instead of doing O(1) work you end up doing O(N) work, so how much
>> slower that is depends on your N.
>> Fwiw, the range of N for web pages I commonly visit is somewhere from
>> 5e2 (e.g. Google homepage) to 3e5 (e.g. the WHATWG HTML specification).
> Thanks for your meaningful reply, Boris. Though, my question was primarily targeted to editors of the Selectors spec itself since they are probably only who know certainly exact reasons and criteria of "fast"/"complete" profile separation.

Boris is one of the major firefox devs, and is part of the reason we
made the profile separation in the first place.

> As for your examples, the estimation I've requested is about a common/average case while the single-page version of the HTML spec if very far from being a common case (in fact, it's even somewhat crazy: it manages to almost freeze my Firefox on my 4-core Intel i7 CPU for several seconds).

That's why it's an upper bound. ^_^

Received on Monday, 9 March 2015 21:39:49 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:52 UTC