From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cagney To: Mark Salter Cc: eliz@is.elta.co.il, gdb-patches@sources.redhat.com Subject: Re: set/show remotestopbits Date: Wed, 26 Sep 2001 11:09:00 -0000 Message-id: <3BB219B8.2090403@cygnus.com> References: <200109261227.f8QCRjr27498@deneb.localdomain> <1858-Wed26Sep2001164321+0300-eliz@is.elta.co.il> <200109261502.f8QF23T27912@deneb.localdomain> <3BB209FF.4050407@cygnus.com> <200109261720.f8QHKWK28267@deneb.localdomain> X-SW-Source: 2001-09/msg00358.html > > I don't know what happened to Don, but I trust I don't want it to happen > to me. Everyone, including me, pulled Don in different directions leading him to implement the same patch N times :-( > I put it in cli-cmds.c because that is where set/show remotebaud lives. I was thinking more of where the command lived within the CLI and its semantics. As to where to put the code well ... Ok, cli/cli-cmds.c contains: > /* If target is open when baud changes, it doesn't take effect until the > next open (I think, not sure). */ > add_show_from_set (add_set_cmd ("remotebaud", no_class, > var_zinteger, (char *) &baud_rate, > "Set baud rate for remote serial I/O.\n\ > This value is used to set the speed of the serial port when debugging\n\ > using remote targets.", &setlist), > &showlist); so it only takes effect when the port is opened - lets call this a feature :-) Things like monitor.c and remote.c then check/set the baud_rate after doing the serial open. The convention has been for commands that affect the serial/parallel port to be called ``set/show remoteXXXX'' eg: set/show remotebaud set/show remotelogfile set/show remotelogbase so set/show remotestopbits certainly follows this convention. We've also recently gained the convention for: set/show remote XXXX to be used for things that relate to the remote targets (using the remote protocol). My first thought is that, the ``serial'' (well serial / parallel / ...) commands should be under: set/show serial XXXXXX and set debug serial rather than ``set/show remote XXXX'' since they are not specific to the remote protocol. Thoughts? I'm pretty easy to convince otherwise. If there is agreement on this, I can churn out the framework needed so that you, Mark, can just drop in ``set serial stopbits''. Andrew