Subj : src/sbbs3/zmodem.c
To   : Git commit to main/sbbs/master
From : Rob Swindell (on Debian Linux)
Date : Fri Aug 21 2026 21:15:30

https://gitlab.synchro.net/main/sbbs/-/commit/36c17d6b1e4c14b79b0c3ccd
Modified Files:
	src/sbbs3/zmodem.c
Log Message:
zmodem: stop stalling a second before the data of every file sent

zmodem_handle_zrpos() purged the receive buffer with a 1-second timeout.
That purge sits on the sender's normal path (every file's first ZRPOS is
just the routine answer to ZFILE), and with nothing to discard, the purge
can only conclude the buffer is empty by letting the read time out. Each
file transferred therefore cost a flat second of silence between its
ZRPOS and its ZDATA, whatever the line speed.

Pass the purge timeout in: 0 for that initial ZRPOS, where anything worth
discarding has already arrived, and 1 for the mid-transfer ZRPOS that
follows an error, where waiting briefly for the rest of the receiver's
retransmit request is the point.

Sending four files over a pipe took 4.01 seconds before and 0.006 after,
to sexyz and to lrzsz's rz alike.

The purge dates to 76d678710e (sheets-20-lone, 2020-11-28).

Reported by Uwe Ohse, who found it comparing sexyz against lrzsz and
zmtx/zmrx over a simulated 115200 bps line: 5.9 KB/s where the others
managed 10.2 KB/s.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
--- SBBSecho 3.37-Linux
 * Origin: Vertrauen - [vert/cvs/bbs].synchro.net (1:103/705)

.