|
|
>>>>> "des" == des small <vonbladet@xxxxxxxxxxx> writes:
des> Lee Sau Dan <danlee@xxxxxxxxxxxxxxxxxxxxxxxxxx> writes:
>> >>>>> "des" == des small <vonbladet@xxxxxxxxxxx> writes:
>>
des> Is there a source control system that handles esoteric things
des> like not having the 'line' as a unit in its diff algorithm?
>> Emac's ediff can refine that to character level diff.
des> What is "character level diff"? What I want is that the diff
des> algorithm ignores semantically irrelevant whitespace in LaTeX
des> sources.
Ediff can take a region of text (derived from the output of plain
diff) and do the diff on that region, not using line as a unit.
BTW, did you know that "diff -B" ignores whitespaces?
>> Why use closed-source, UNDOCUMENTED, proprietary formats?
des> I mostly don't.
That question is addressed to everyone, not just you specifically.
>> XML files are structure as trees, and tree-diffing is not a
>> well defined problem. i.e. When you say "give me a diff of
>> these 2 trees", there is no single answer. For text, we can
>> define the "optimal" answer to be the one with minimal editing
>> distance. And text 'diff' tools are based on this. The
>> optimal answer gives results that we want. However, for
>> tree-diff's, it is difficult even to define what is "optimal".
>> What should be "optimal" varies from application to
>> application.
des> So what you're saying is I'd still be screwed with ODF for
des> the time being; double whoop.
I mean you're asking for something (XML diff) to be done, but that
something is not described (please define what is an "optimal" tree
diff) clear enough for computers to handle in a way that would satisfy
you. Remember, computers are very stupid and can't read your mind.
--
Lee Sau Dan +Z05biGVm- +AH4-{@nJX6X+AH4-}
E-mail: danlee@xxxxxxxxxxxxxxxxxxxxxxxxxx
Home page: http://www.informatik.uni-freiburg.de/+AH4-danlee
|
|