Dylan Simon
Dylan Simon
Should we set common environment variables like `OMP_NUM_THREADS`, `MKL_NUM_THREADS`, etc. to the appropriate `-c` value? I've noticed a number of disbatch tasks each trying to use all the cores, so...
When a task fails, we currently continue. Allow different behaviors, such as exiting immediately, or exiting after running tasks finish, or just propagating exit code to disBatch somehow.
Should keep the status log, but can suppress other files if requested.
this would allow users to mix tasks with different threading requirements it would have to be an integer multiple of the current setting, and checked against the minimum workers/host
(This is with ganesha 3.5-ibm071.07, which I think is roughly ntirpc v1.8.0, but I wanted to mention it here because it looks like the current version still mishandles reference counts...
When scrolling multiple lines an a time when part of the window is obscured, so that some lines are moved and some are redrawn, there is corruption. Observed in copy...
Starting with 1.8, new xtmux windows sometimes end up sized 80x24, perhaps because of an out-of-order resize.
xtmux.c is too big. Some things can be redistributed other places, other things should be put in xtmux.h or something, and associated structs should be renamed.
In copy mode, every time the selection is updated (e.g., by moving the mouse), it redraws the whole screen, which is flickery and slow. Either window-copy could be smarter about...