Hi,
while doing the re-implementation of unified diff for svn I
noticed a
few possible problems with our current approach to diff():
- should we analyze the diff (in case of unified diff) and
extract all
paths from that and put their content into the leftTexts()
list of
locations? I'm doing that at the moment, however this
means you can't
take the leftLocation and apply the diff to it to get the
result
- if we allow recursive diffs, how do we create leftText and
rightText?
I mean we have directories here and I doubt any VCS gives
you a full
content-string for a whole dir. For unified diffs one can
extract the
left and right location from the diff, i.e. only have the
changed
locations, but for the raw diff this would mean
recursively fetching a
whole VCS structure...
To conclude: I'm not sure how, but I think we need to revise
our
approach to diff(). Ideas?
Andreas
--
Good day for overcoming obstacles. Try a steeplechase.
_______________________________________________
KDevelop-devel mailing list
KDevelop-devel kdevelop.org
https://barney.cs.uni-potsdam.de/mailman/listinf
o/kdevelop-devel
|