On 22-07-03 12:24am, Christian Weisgerber wrote:
> Omar Polo:
> 
> > The last patch from Mark reminded me that I thought some time ago to add
> > a "-w" flag for got patch to ignore white spaces.  It's conceptually
> > similar to patch(1) -l/--ignore-whitespace or git-apply(1)
> > --ignore-whitespace.

Firstly, I think this option is great. Thanks, op :)

> > The diff does exactly what one would expect.
> 
> Well, I'm dense.  What does "ignore whitespace" mean exactly?
> Note that diff(1) has two ways to do this, -b and -w.

I had the same ambiguity when reading the docs for this too.
Also, regarding "Ignore whitespaces in context lines."; how is "context
lines" defined?

For instance, -w made my diff apply cleanly, but the whitespace was on
a changed line.

> If I understand linecmp() correctly, it's like diff -w?
> 
> Meanwhile patch -l seems to be more like diff -b?

I like Fossil's semantics for this:

  -w|--ignore-all-space       Ignore white space when comparing lines
  -Z|--ignore-trailing-space  Ignore changes to end-of-line whitespace

but I think -w and -W might be better flags than -w and -Z.

I'm not sure how I feel about this being the default though. I admit,
it's nice not having failed patches, but as far as I'm aware, no other
patch implementations do it by default so it might fail the Rule of
Least Surprise. And--ironically coming from the person who sent out the
mangled diff that instigated this--it shouldn't happen often. I can't
even remember the last time I did it. Plus, with the:

"@@ -295,7 +305,7 @@ hunk contains mangled whitespace"

output, users can then rerun with -w.

That said, I really am ambivalent and can't decide which way I fall on
this.

-- 
Mark Jamsek <fnc.bsdbox.org>
GPG: F2FF 13DE 6A06 C471 CA80  E6E2 2930 DC66 86EE CF68