Re: Changes to grep/src/grep.c

[Charles Levert]

* On Thursday 2005-07-07 at 10:43:42 +0100, Julian Foad wrote:
> Charles Levert wrote:
> >
> >Done.  Hopefully it will be considered at one
> >point and won't disappear into oblivion.
> 
> Many thanks.  Note that I wasn't objecting to the idea of that feature - I 
> can see how something like it might be useful.  However, before adding code 
> for a new feature you should explain exactly what the feature is intended 
> to be, and why, and get some approval.  Please feel free to keep developing 
> the idea and propose a new feature if you want to.

Note that I consider that the rest of the
code, the basic framework in particular, is now
better because of this.  That is, because of the
generalization of what boolean capabilities can
do and what they need from the framework.

For instance, older versions had small unsigned
values cast as pointers to have "magical"
meaning.  Gone, and good.  The ability to call
a function for each specified capability is much
cleaner, better, and generic; that remains.

The xm capability forced me to perform true
hierarchical nesting which real SGR sequences
cannot use (because they must re-initialize
the colors after an SGR_END).  I removed this
code for now, but hopefully the fact that this
will remain as a one-time CVS revision will
be there to document what needs to be done to
support both.  That's a useful thing to be able
to understand, IMHO, regardless of the exact
<grep:foo> output syntax that was there and
that may likely be what first caught your eye
about this.