Add VT sequence: DECSCLM - Scrolling Mode

[christianparpart]

We currently only implement fast mode, but it might make sense to also implement smooth (slow) mode.
This mode only affects adding new lines at a slow rate, but - in modern age - we could correlate that feature with smooth scrolling (of new lines to be added), see #145.

One open question I am having is how long the linefeed should wait until resuming terminal I/O processing.

How to implement

One design idea I am having in my mind is, to actually suspend processing for N msecs on received linefeeds, if slow scrolling mode is enabled. This should effectively also suspend reading from the TTY to avoid flooding the internal buffers.

References

• https://vt100.net/docs/vt510-rm/DECSCLM.html

[j4james]

One open question I am having is how long the linefeed should wait until resuming terminal I/O processing.

On the early DEC terminals (VT100, VT220, VT240, and VT340) the smooth scrolling speed was 6 lines per second. The VT340 apparently also had an option in the settings that could adjust the speed (choosing between 3, 6, and 12).

On level 5 terminals (like the VT510, VT520, and VT525) the default smooth scrolling speed was 9 lines per second, but there was also an escape sequence to adjust that to 18 (See DECSSCLS).

I think the choice was largely dictated by the hardware specs. For example, the VT100 had a 60Hz refresh rate with a cell height of 10px, so scrolling one pixel per refresh gives you 60/10 lines per second. For the VT510, you've got 72Hz and a 16px cell height, so scrolling two pixels per refresh gives you 72*2/16 lines per second.

[Yaraslaut]

What will happen if I have a stream that contains more lines that I want to have in the history and then enable slow scrolling? as an example find /

Edit:
Oh right, i see that the idea is to to stop I/O, will it work with extensive logging from the process?

[christianparpart]

will it work with extensive logging from the process?

yes, because the terminal I/O thread is slowed down due to the artificially injected delay upon each LF to be executed.