Re: backwards compatibility, was Re: [PATCH v1 1/3] Introduce config variable "diff.primer"

From: Jeff King <peff@peff.net>
Date: Jan 26, 2009, 18:48 UTC
Message-ID: <20090126184829.GA27543@coredump.intra.peff.net>
In-Reply-To: <76718490901260729m21ba140dke157d1d461aed2d5@mail.gmail.com>

On Mon, Jan 26, 2009 at 10:29:17AM -0500, Jay Soffian wrote:

Show 11 quoted lines

> Aside, AIX has commands that are run both directly or via smit (a
> curses-based interface). When smit calls the commands, it passes a
> switch to let said commands know that they are being run from smit.
> e.g.:
> 
>        -J
>             This flag is used when the installp command is executed from the
>             System Management Interface Tool (SMIT) menus.
> 
> Perhaps adding such a concept to those git commands which can be used
> in both porcelain and plumbing contexts would be useful for git.

Sure, I think that is one of many possible ways that we could differentiate between confusing plumbing and porcelain; another is splitting functionality into two similar commands, one of which is plumbing and one of which is porcelain.

The real problem with plans like that, though, is that there are _already_ scripts in the wild that don't understand "-J" (or whatever). My impression from your description above is that "-J" means "don't use fancy features, because we're being called from the menus". And you really want the opposite, which is that scripts opt _in_ to fancy features, not _out_.

But then you have that problem that the _user_ is stuck specifying "OK, turn on fancy features." And I don't relish the thought of typing "git diff -J" every time. :)

-Peff