Experimental Physics and
Industrial Control System

"J. Lewis Muir via Tech-talk" <tech-talk at aps.anl.gov> · Thu, 18 Jun 2020 17:09:02 -0500

On 06/18, Johnson, Andrew N. wrote:
> On Jun 18, 2020, at 2:24 PM, J. Lewis Muir <jlmuir at imca-cat.org> wrote:
> > If you're asking whether tmux can log, the answer is yes, using the
> > pipe-pane command.
> 
> No, I’m asking what tmux does if you run an IOC in it that generates messages occasionally, then use the ability to look back in its scroll-back buffer and leave your window scrolled to the very beginning of the buffer for a long time after the buffer has reached its size limit. Does it eventually freeze the IOC by imposing flow-control on the stdout/stderr stream from the IOC?

Ooh, that wouldn't be good. :-)  Didn't know about such an issue.  I
looked, and it appears that tmux did have an issue like that, but it was
fixed with commit bee3e3e in tmux 2.3 (released on Sep 29, 2016):

  https://github.com/tmux/tmux/commit/bee3e3e28d04a237b1013b1dd3d36cddcd326891

The fix adds a timer that will automatically exit out of copy mode
(i.e., the scrollback buffer) after three minutes of inactivity and if
there is pending output.

I confirmed that the timer works in tmux 3.1b on RHEL 7.

Strangely, the timer does *not* work in tmux 3.1b on macOS Catalina
10.15.5.  Not sure why.  I'll have to investigate.

Interestingly, there's commit 77b827f that removes the timer logic and
instead makes a copy of the pane's history for copy mode:

  https://github.com/tmux/tmux/commit/77b827f879e4dc296996b3f4345c38a7a57caf48

But that commit does not seem to have made it into a release yet.

Lewis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System