On Sat, 2010-03-27, Stefan Küng wrote:
> On 27.03.2010 15:23, Stefan Sperling wrote:
> > On Fri, Mar 26, 2010 at 07:57:09PM +0100, Stefan Küng wrote:
> >> On 26.03.2010 18:17, Stefan Sperling wrote:
> >>> Can you check if the current API matches your requirements?
> >>> See subversion/tests/libsvn_client/client-test.c for a trivial example.
> >>
> >> Looks very good! Thanks a lot for implementing this.
> >
> > Great! You're welcome.
> >
> >> Another thought, not sure if it would make sense or not:
> >> would an option to ignore whitespace changes make sense?
> >
> > Yes, I guess we could easily make the line-matching ignore whitespace.
> > Matching is currently done using a simple strcmp(), so it's fairly naive.
> >
> > Please file an enhancement issue, with undefined schedule (if someone
> > wants to pick this up before 1.7.0, that's fine -- but it's not a
> > critical feature).
> >
> > Ignoring whitespace would help dealing with diffs messed up by e.g. gmail.
> > Though I've seen gmail use some magic UTF-8 sequences for whitespace,
> > which is really annoying and which a naive approach like ignoring
> > single-byte characters like ' ', '\t', etc. may not cope with.
>
> Filed as issue 3610:
> http://subversion.tigris.org/issues/show_bug.cgi?id=3610
Can I recommend you add more detail to the request. "Ignore whitespace
changes" could be interpreted in many ways, and defining what exactly
you mean is a big part of the issue.
For example, one change that some mail clients make is to remove a space
from the beginning of every line that starts with a space (or starts
with two spaces?) so that (using "_" to represent a space):
__context-line
__context-line
__context-line
+_plus-line
-_minus-line
__context-line
__context-line
__context-line
has turned into
_context-line
_context-line
_context-line
+_plus-line
-_minus-line
_context-line
_context-line
_context-line
when the mail is received.
If the extension we want is just for that, then
"Allow a context line to match a real line that has an extra space at
the beginning"
is a sufficient definition, whereas
"Allow any sequence of whitespace (including an empty line) in the
context lines and minus lines to match any sequence of whitespace
(including an empty line) in the target file"
is a very different requirement.
And/or do you want
"When a hunk performs only whitespace changes, ignore that hunk"
?
- Julian
Received on 2010-03-29 15:08:41 CEST