Wedge

Public area => The Pub => Off-topic => Topic started by: MultiformeIngegno on April 5th, 2013, 05:58 PM

Title: Zopfli compression
Post by: MultiformeIngegno on April 5th, 2013, 05:58 PM
http://josephscott.org/archives/2013/03/zopfli-compression/
Title: Re: Zopfli compression
Post by: Arantor on April 5th, 2013, 07:25 PM
Yes, because increasing server load by a factor of 100 (for stuff that could be updated semi frequently) could be a good thing.

It has uses (and it's old news), but it doesn't really apply to us much.
Title: Re: Zopfli compression
Post by: MultiformeIngegno on April 5th, 2013, 07:32 PM
That's why I didn't post in Features :P
Title: Re: Zopfli compression
Post by: Arantor on April 5th, 2013, 07:35 PM
Heh, well it is interesting in itself and there are times and ways it will be useful but I think those are rather few and far between :(
Title: Re: Zopfli compression
Post by: Dragooon on April 5th, 2013, 07:39 PM
I can see its use in CDNs where the content is fairly static.
Title: Re: Zopfli compression
Post by: Arantor on April 5th, 2013, 07:41 PM
But that's something that the CDN would typically manage transparently, no?
Title: Re: Zopfli compression
Post by: Nao on April 6th, 2013, 01:42 PM
Yes of course, and that's really good news for jQuery distribution, among other things... Another kilobyte saved :)

There are two PHP implementations so far:
https://github.com/kjdev/php-ext-zopfli
https://github.com/clickalicious/php_zopfli

May the best one win!
Couldn't find a compiled DLL for local tests though, so I guess it'll have to wait... (And yes, I'm not against adding support for zopfli in Wess, as long as low-iteration passes have similar performance to gzip -9 and still save more bytes.)
Title: Re: Zopfli compression
Post by: MultiformeIngegno on April 6th, 2013, 03:29 PM
And now that you're all minified, stretched and compressed take a look at this :P http://www.fess.me/
Title: Re: Zopfli compression
Post by: Arantor on April 6th, 2013, 03:39 PM
Quote
Couldn't find a compiled DLL for local tests though, so I guess it'll have to wait... (And yes, I'm not against adding support for zopfli in Wess, as long as low-iteration passes have similar performance to gzip -9 and still save more bytes.)
That's the point, it's at least one order of magnitude slower than gzip -9, if not nearer two orders of magnitude slower.