> From: Richard Stallman <[email protected]> > Cc: [email protected], [email protected], [email protected] > Date: Fri, 09 Jan 2026 10:49:03 -0500 > > GDB is the GNU DeBugger, so good Emacs support for GDB is very > important. It is useful to make Emacs work well with other debuggers, > but making it work well with GDB has to be the priority. As long as > GUD supports GDB significantly better than Realgud, GUD is more > important for us than Realgud. > > It sounds like making Realgud support GDB via MI, and then replacing > GUD with Realgud, could be a significant improvement in GNU Emacs. I > hope people will work on that. > > Using the GUD code for MI as a model, and maybe copying parts, > it should not be terribly hard.
We have already done that, years ago. There's the gdb-mi.el package, which uses MI to communicate with GDB, and "M-x gdb" by default uses that. This has been our default since Emacs 24.1. GUD now stays in Emacs mainly for the benefit of debuggers other than GDB, and for those few Emacs users who still prefer "M-x gud-gdb" and its CLI-based interface with GDB. So if we are to consider bringing Realgud into the core, I see no real benefits in adding MI support to it, unless we also want to replace gdb-mi.el (in which case Realgud will need to support the additional UI features gdb-mi.el provides, if it doesn't already). If we are keeping gdb-mi.el and replacing gud.el, a small part of the current gud.el will need to be left behind, because gdb-mi.el uses some of its facilities.
