value be used. An empty/blank value means '0' (no bit flags).
Resolve a couple warnings:
'lp' is assigned a value that is never used

This fixes the loading CTRL-A codes from *.msg files reported by Marius.

cryptlib stuff may do Bad Things.

from mailslot (part of fix for upload to FDSZ).
Bug-fix: don't potentially overwrite the bounds of the rx ring buffer.
Better input_thread debug logging.
Bug-fix: VDD_STATUS could return garbage for input full/size if GetMailslotInfo
fails (which it appears is doing) - which could be an issue for FOSSIL apps
which rely on function 5.
Specify unlimited-size msgs for receive (input) mailslot (part of fix for FDSZ
upload).

seconds, just continue waiting (and logging warnings).

could stay in binary mode.

Allow transfer abort callback to ignore mouse keys (and each extended keys)

movements.

(thanks!)

remove frame.getoff() method, add frame.cursor, frame.offset properties, use "y" based matrix for frame data. use self-validating cursor and offset objects for position. update documentation.
ToDo: punch deuce

therefore the current line ends with a hard CR), allow CR and LF to be
detected as whitespace as well.

This is simply removing code though, so the expectation that I wrote that
code for a reason is haunting me.

Refresh every 30 seconds rather than every 10.  Should result in decreased rectal pain for webservers and browsers alike.

TODO: figure out why syncterm screws up line wrapping in an 80-wide frame
TODO: blame deuce

restrict scrolling based on v_scroll/h_scroll settings
update shitting documentation
attempt open/close regardless of frame state

add getoff() method (ooo yeah...)
add v_scroll, h_scroll, scrollbars_visible settings (*TODO*)
add lf_strict setting (force linefeed on crlf-terminated putmsg())
no longer force open() when load()ing a graphic