Not someone to give up, I've got winlirc running under win2k.
I've tried using the Tx with Break/Space, and RTS, but all of these
are still slow, probably due to Win2k doing some security checking
on the DTR on/off command. It is about 10 times slower than win2k,
which is just barely ok (the cycle on 38Khz is about 13 usec,
and win98 I think take about 4usec for on/off on 1Ghz AMD cpu)
I've used a hack known as userport.zip which allows a win2k process
to use direct amm in/out instructions.
I've added code for asm-level DTR on/off, using an optional "/P" parm,
and it works fine.
If WinLirc "offical" version want this update, let me know.
Note that userport produces a security hole, and is basically
installed "for all programs"...
I've also tried winio.dll from internals.com which is much safer
in a sense (its own dll load its own kernel-privildge device)
but the switching between user & kernel mode apparently take too long.
I've also written IRSERV which is an HTTP mini-server that allows
"GET /trans/x/y HTTP/1.1" to transmit device x, key y. This allows
me to easily control my TV cable box, near my desktop PC, via
irserv from a notebook anywhere in the house (wireless ethernet).
Sorry, can't release irserv as it uses a proprietary lib.
Even so, all is NOT well.
Winlirc just isn't as good as a dedicated hardware. I'm running
Real's realserver/producer on the desktop, along side winlirc/irserv,
and as the cpu is 60-70% used, winlirc is unable to consistently
transmit commands. Some times it would work, sometimes not. If the
cpu load is reduced to near-idle, winlirc works nearly-perfect.
I'm now looking again into buying a one4all remote and the JP1
project for transmit. Any other/better ideas or a dedicated winlirc
hardware I could use?