Comments on: Run-length encoding (part I)

By: Daniel Lemire

Daniel Lemire — Thu, 26 Nov 2009 12:13:33 +0000

@Kevembuangga Let me put it differently. Could you implement the a scheme like the one you described?

By: Kevembuangga

Kevembuangga — Thu, 26 Nov 2009 07:43:13 +0000

P.S. To explain my (lack of) motivation, since I do not personally enjoy the hairsplitting about nitty gritty details and umphteenth decimals which show up in “academic” research I don’t feel it’s worth the effort to elaborate on this.
I am not even sure it is worth your efforts, it may lack “nice properties” to write about, yet it was effective in a commercial product.

By: Kevembuangga

Kevembuangga — Thu, 26 Nov 2009 07:31:07 +0000

I’ll gladly study any form of RLE if I have proper documentation.

I am afraid there isn’t any documentation.
As I told you this is something I stumbled upon when peeking at reverse engineering of a compression utility.
This wasn’t academic, only engineering, sorry

By: Daniel Lemire

Daniel Lemire — Wed, 25 Nov 2009 12:27:02 +0000

@Kevembuangga I’m interested in RLE because it is a fundamental idea. I’ll gladly study any form of RLE if I have proper documentation.

I could try to guess what iterated hierarchical RLE with a fancy encoding means, but that is not a very interesting game.

By: Parand

Parand — Wed, 25 Nov 2009 08:42:47 +0000

Thanks Daniel for the beautifully concise explanation of RLE. Makes me wonder how it can seem complex in other explanations.

By: Kevembuangga

Kevembuangga — Wed, 25 Nov 2009 07:36:05 +0000

With this “obsessive” interest in RLE may be you should think again about the compression scheme we talked about in emails on 19 Sep 2009 and which you didn’t seem to grok:
It’s actually iterated hierarchical RLE with a fancy encoding of (what serves as) “counters”.