The Random Thread

mvoorhis said:
Does anyone know the historical basis for the ^T ^O keys? This function has to be written into various user programs and it takes effort to maintain these, etc. I'm just curious to know more about the reasoning behind these? I find them convenient at times but they seem specific to FreeBSD and perhaps to older UNIX.
Click to expand...

I like them a lot. Getting a progress report without having to ask for them at startup is convenient.

Only problem is that some people are tempted to use printf in the signal handler, which is not legal.
 
mvoorhis said:
In FreeBSD there is a convention where one can type ^T or ^O (Control-T, Control-O) in the midst of the run of various programs (will use DD as an example). ^T will request some sort of runtime statistics from the program and ^O will temporarily suspend output to the terminal.

These two are in addition to the much more widely used ^S and ^Q, ^U, etc. As far as I can tell only FreeBSD still supports ^T and ^O. These keys don't do anything in any Linux I've used.

Does anyone know the historical basis for the ^T ^O keys? This function has to be written into various user programs and it takes effort to maintain these, etc. I'm just curious to know more about the reasoning behind these? I find them convenient at times but they seem specific to FreeBSD and perhaps to older UNIX.
Click to expand...
Check out pages 7 (^O) and 20 (^T) of the following 1975 DECsystem-10 manual. DECsystem-10 was not Unix (as far as I know), and yet it incorporated these commands, which seem to be part of an earlier set of terminal conventions. Anyway, the document is what's cooler.

 
mvoorhis said:
The keys generate special interrupts? Via the shell?
Click to expand...

They are signals and they are executed as signal handlers in the target program.

Signal handlers have restrictions on what code can be executed in them. One restriction is that they cannot use malloc(3), which precludes use of printf. Without printf it is more difficult to print things like percentages.
 
cracauer@ said:
[...] Signal handlers have restrictions on what code can be executed in them. One restriction is that they cannot use malloc(3) [...]
Click to expand...
This is the part I didn't know. Thank you!

And thanks also to Alfredo with the DECSystem-10 (TOPS-10?) information. I remember using ^T and ^O in TOPS-20 but that's old knowledge and I didn't remember it until your reminder. :)
 
I judge my contributions to these forums poor. They are generally unserious and lack technical depth. I'm an exotic presence more than anything else: faintly tolerated and quite tiresome.
 
I think it's very commendable not to have a linkedin account. I never had one either. I don't even know what people use linkedin for. I don't even know how to spell it right; I have to check every time how to spell it.

Your anecdote is very revealing. In some circles, deviating even a little bit from the "norm" is received with shock. "Struggle to process this revelation" is a very good (and funny) description of said fact.
 
AlfredoLlaquet said:
I wonder why T-Aoki doesn't have the developer "sash." He clearly is a developer of a very popular component of the system.
Click to expand...
I think it is displayed if the registered email address of the account is *@freebsd.org. This is basically assigned to committers (some exceptions like non-developer admins of infrastructures like builder clusters).
 
You must log in or register to reply here.
Back
Top