cdiff should detect if there is a terminal

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

Olav Vitters wrote:
> Public bug reported:
>
> I'd like to alias diff to cdiff. This so I do not have to remember if I
> should run cdiff or diff. However, cdiff doesn't automatically disable
> its colouring when the output goes to a terminal, preventing me from
> aliasing it.

However, I pipe cdiff through less on occasion, and less is not a
terminal, so that would cause color to be disabled.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIXoSF0F+nu1YWqI0RAvGmAJ9vCV5D8AMYGaRW8zyt9wxN9Y4dWwCfbbBv
ZN94YSVXl/EwUG8nYqAW6Os=
=waTN
-----END PGP SIGNATURE-----

Perhaps an --color option like grep? With 'never' (bit strange), 'always', 'auto'. Ideally I'd even like diff to automatically do cdiff --colour=auto :)

I like the way "git diff" works: it sends color codes if the output is a terminal, but has also the options --color and --no-color to force it.

This doesn't seem fixed. With bzr.dev 4359 and bzrtools 705 (both from right now), „bzr cdiff >file; cat file“ shows the file in colours. Same with „|less“. GNU/Linux Debian and python 2.5.4

The problem is that --color=always is the default, instead of --color=auto, which is more a reasonable and extended behaviour.

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

Daniel Clemente wrote:
> The problem is that --color=always is the default, instead of
> --color=auto, which is more a reasonable and extended behaviour.

color=auto will disable color when cdiff is piped to less -R, which is
broken. It is entirely appropriate for a program designed to emit
colorized diffs to default to emitting colorized diffs.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkoMMCUACgkQ0F+nu1YWqI0MPgCfS6/yvD8sJrlqO701oHMbKviP
TA8An1S5sDzDO5YjNQ18Humnl87beWit
=TPRf
-----END PGP SIGNATURE-----