Tuesday, April 15, 2008

My Opinion: Let's Not Use RAR

I had to do some research into compression algorithms recently, because we had to squeeze the global scenery onto fewer DVDs for retail distribution.  We did this mostly by completely filling the DVDs, but we also had to use 7-zip compression to get about a 10% improvement in compression ratios.

DSFs are not the best test of compression efficiency because the format has been organized to help algorithms like zip compress them - the improvement with 7-zip and RAR was a lot less than you'd get with, say, a text file.

Anyway, my point here is: let's not use RAR - it's the new GIF.  Every now and then a file format comes along with some kind of restriction that keeps everyone from doing everything with it.  In GIF's case, you had to buy the right to create GIFs, and in the case of RAR you have to buy the right to compress RARs.

I think that having these kinds of entanglements in fundamental low level file formats (like how do we compress our data or save our images) is really bad for the software community as a whole; it balkanizes raw materials.  And file formats stick around for a long time - even though GIF is made obsolete by PNG you'll still find them all over the web.

The lure of RAR is of course higher compression ratios than zip.  But 7-zip can do the same thing, and unlike RAR, has the potential to be completely free, which means it can be completely ubiquitous.

Macintosh users understand the problem here: for the longest time "StuffIt" archives were the standard way to compress data on the Macintosh.  The file format was proprietary, so you couldn't even make your own program work directly with StuffIt archives.  Now that zip has taken over on the Mac, getting data between Mac and Win is easy - you can just zip something using the operating system and send it to all your friends.

Let's not go back into the "bad old days" of proprietary utilities and a lack of integration with regular apps.  I say: if you can stand to use zip or bzip instead of RAR, vote for what's open and has a future, not what is slightly better now but will just be a pain in the ass in three years.

No comments: