This is a tough issue as I spend much of my day in the command-line world and I do not disagree with your basic statement about needing a GUI, these days even though it is more of a ball and chain than a helpful tool. It's like a sore knee or a backache. Nature usually fixes those in time, but the GUI devours resources and there is always that one last problem that keeps it from working right. I have both a Macintosh for the GUI and I use speakup under Debian wheezy with lynx and nmh under FreeBSD for mail. This last bit has nothing to do with screen readers but mh or the package now known as nmh breaks the email process in to small modules that allow one to automate different parts of the mail process. Part of my job is building automation that sends messages to others when various things happen so the use of nmh is a choice. I have yet to get orca working on any system I use or have access to. One such system is a Pentium4 running at 2.7 GHZ and there is a gigabyte of RAM sitting there but there is something in the BIOS that seems to know when I want to install the latest ubuntu or Debian that might open up the world of gnome and orca and the system figures out some clever way to fail. By the way, speakup works beautifully on this system in a command line console but The only time I ever heard orca talk was on an obsolete version of ubuntu 9.0 which played for sometimes an hour or so and sometimes a few seconds and then would crash. You are correct in that basically, the speech process needs to be separate from just about everything except the power supply in order to hear the system start up from black. A Unix kernel is the master process and everything else that happens on your system is spawned as a subprocess of the master. Would it be possible to have a kernel equipped with speakup spawn the rest of one's system as if it was a virtual system? That could take care of the I/O. I used a hardware speech synthesizor for about 20 years along with Kermit and DOS and a screen reader I wrote to terminate and stay resident in MS-DOS so all my Unix boxes were originally configured for a RS-232 console. That was back when mother boards had RS-232 ports. You've really got to separate the speech or Braille output from the rest or it will always bite you. Speakup should go in a sort of pre-kernel and that would let you operate the real system in single-user mode, listen to kernel messages and do all those things we should do if we are to call ourselves Unix administrators. Martin "John G. Heim" writes: > I totally disagree. Speakup has little purpose except for the fact that it > runs in kernel space. First of all, there are other screen readers for > user > space. And you really need a GUI these days. I suppose there are people > using speakup all day every day. Mutt for email, lynx or edbrowse for the > web. > But I'm sure the vast majority of linux users use orca for every day > tasks. > > > > The most important feature for speakup is to bail you out when you are > really in trouble because your server is down. I don't know what you do > for > a living but I do systems admin and I cannot live without speakup in > kernel > space. About the only thing that I can think of that is equivalent to > simply plugging in a hardware synth and getting boot messages would be > setting up something like a Raspberry Pie to boot into kermit and display > serial console messages. But it wouldn't be the same because you'd need a > keyboard for the RPI. I don't know -- when a server is down, the last > thing > I want to do is mess with all that stuff. I just want to plug in the > hardware speech synth and press the print screen key.
