Talk:Delta encoding

From Wikipedia, the free encyclopedia

Contents

[edit] C code is wrong

I think the C code is not correct

You are correct. If you run the example in the text trough this algorightm (2, 4, 6, 9, 7), the encoded becomes: 2, 2, 4, 5, 2
The decode then becomes: 2, 4, 8, 13, 15
The decoder is correct (checked with encoded data 2, 2, 2, 3, -2): 2, 4, 6, 9, 7 (which is the right awnser).
I will correct the encoder after I add this comment. ShadowLord 09:14, 7 Apr 2005 (UTC)

[edit] Generalisation of delta encoding

Could this article be expanded to a more general view of delta encoding as encoding only the differences between things? A mention of diff and similar might be quite good. porges 09:38, Apr 7, 2005 (UTC)

I've changed the text to link to delta compression article, which have more direct relation to revision control in general and diff in particular. 217.26.163.26 07:23, 4 May 2005 (UTC)

[edit] Error control

Error control has different meaning in this article - it's used to find differences from other version of the same file without transfering it, not to detect corrupted data. This is the way how rsync/zsync use it. Please, don't change this meaning! 217.26.163.26 07:06, 4 May 2005 (UTC)


[edit] symmetric delta

The article currently states

A symmetric delta can be expressed as \Delta(v_1, v_2) = (v_1 \backslash v_2) \cup (v_2 \backslash v_1), where v1 and v2 represent two successive versions.

OK, that is a pretty formula, but what does it mean? (What does the "\" mean in this context?)

I'm guessing it means something like

Given two versions v1 and v2, we can compute "the delta" between them.
A directed delta tells us, once we know v1, how to generate v2 -- but if we only know v2 and the directed delta, it's generally impossible to recover v1.
A symmetric delta can go both ways -- if we know v1, it tells us how to generate v2. But if we only know v2, that same symmetric delta tells us how to generate v1.

Is that all it means ? Is it OK if I replace the mysterious formula with this English-language description? --DavidCary 03:19, 6 October 2005 (UTC)


If I read the formula correctly, it means the same as \Delta_s(v_1, v_2) = (v_1 \cup v_2) \backslash (v_1 \cap v_1), which I can read easier in English: the symmetric delta of two versions is the elements of each version without the elements common to both.

I'm guessing that directed delta could be defined by \Delta_d(v_1, v_2) = v_2 \backslash v_1, and would read: the elements of second version without the elements of the first.

I don't feel that the formula adds much to the article. --Eruionnyron 19:10, 9 October 2005 (UTC)

[edit] Potential sources for new article

This article is crappy. I'll rewrite it, but having a new list of sources would be quite helpful. If anyone wants to chip in I'd be very pleased.

[edit] What??

"Perhaps the simplest example is storing values of bytes as differences (deltas) between sequential values, rather than the values themselves. So, instead of 2, 4, 6, 9, 7, we would store 2, 2, 2, 3, -2."

This wouldn't save any space at all. Is this actually used to save computer data file revisions? I highly doubt it. We already have an article for delta modulation, as used in audio (and similar to what is used in video). This article is supposed to be about saving revisions of large chunks of data, right? Involving the unix diff utility? Not about the difference between subsequent samples within a chunk of data. — Omegatron 20:55, 18 May 2007 (UTC)