Re: ignoring whitespace in rdiff

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Clendenan, Dave <address@hidden> writes:

> cvs diff takes '-bw' to ignore whitespace.  cvs rdiff (at least in the
> versions I have) doesn't have an option for this.  
> 
> Can someone please explain this to me?

See below.

> Does the 'feature release' stream have this feature?

No.

> We're developing on windows, and some of our developers keep switching
> linefeed types, which makes rdiff a bit useless. I may have to resort
> to a linebreak-conversion on checkin, but I don't like the idea of
> automated chenges to entire files like that...

Line endings should be properly handled by the client already.


The general topic of additional flags to 'cvs rdiff' (aka 'cvs patch')
came up here:

  https://ccvs.cvshome.org/issues/show_bug.cgi?id=50

and Larry wrote:

|  -- Additional comments from Larry Jones Tue May 7 15:01:06 -0700 2002 --
| 
| "cvs diff" supports nearly all of the GNU diff options but "cvs patch"
| (aka "rdiff") hardly supports any of them because it was designed to
| make patches, not to be a generic diff command.  Given that, I don't
| see any justification for adding "-p" to the miniscule set of options
| that "cvs patch" accepts unless we ignore the original design and make
| it into a generic diff command, which is problematic because some of
| the existing patch options conflict with diff options.

and Derek wrote:

|  -- Additional comments from Derek Price Mon May 20 07:45:29 -0700 2002 --
| 
| Except for breaking backwards compatibility, I'm not sure I agree with
| Larry's argument.  Isn't a straight `diff' the traditional way of
| creating patches when CVS isn't involved?  Also, making `cvs rdiff' a
| remote version of `cvs diff' seems to make sense since that seems to
| be consistent with the operation of the rest of the CVS commands.
| 
| Closing this anyhow until someone with the time to work on it comes
| along and reopens it.

Issue 176 ( https://ccvs.cvshome.org/issues/show_bug.cgi?id=176 )
also requests the -p option for rdiff.

Your request to add -b (ignore changes in amount of white space) and -w
(ignore horizontal white space when comparing lines) are just the latest
in the debate of how to deal with passing all of the possible diff options
to the rdiff subsystem in a sane manner given that we have many options
("D:RV:cflr:stu") to rdiff and many of them conflict with 'diff'
("D:flrst").

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)

iD8DBQFBfoJV3x41pRYZE/gRAstuAJkBM0aL1Ucwsc0Gf5Us07ZZ7QHp6QCdH16k
JlENv9iAo8The/P9pGe+J0w=
=dsrQ
-----END PGP SIGNATURE-----