Re: [PATCH] find-file-noselect-1

[Nick Roberts]

 > BTW, I firmly vote against any such change (even if done via
 > find-file-hook): gdba.el should be trying to be more robust (in the face of
 > unexpected or missing output from gdb), not more featureful.

Currently, in a debug session with gdb-ui, the first breakpoint has to be set
either with a global binding e.g C-x SPC or through the GUD buffer. No other
debugger, that I know of, puts such an unintuitive constraint on the user.
This patch makes the fringe available for setting breakpoints at the start
of the session. I wouldn't call that a adding new feature, but rather making
an existing one more sound.
 
 > Otherwise, we run the risk that we'll need to make gdb's default
 > "gdb --fullname" rather than "gdb --annotate=3" if too many bug-reports come
 > in and we can't fix them in time.

Not many bug reports have come in, probably because not many people have usd
gdb-ui, but thats a separate risk to the one of using find-file-hook. Anyway,
lets wait until too many bug-reports do come in before talking about changing
the default.

If a user doesn't use GDB, then there's no risk because the hook doesn't get
called. If a user does use GDB, and finds a problem with gdb-ui, then he's
probably competent enough to set the default to "gdb --fullname" where the
hook does nothing.

GDB isn't just another application, its part of the GNU Project and, as a
debugger, intimately linked with the editor (I think it even used to be
distributed on the same tape as Emacs). I think its important that these two
programs should be made to work well with each other and that, in the
long term, the benfeit is worth the risk.

Nick