whitney@think.COM (David Whitney) (06/30/88)
Don Elton writes: >TIC (Talk is Cheap) uses the standard ProDOS interrupt vectors and a 256 byte >interrupt buffer for irq handling. When I use 19,200 baud for transfers it >keeps up with the max speed I can get an Apple to transmit at which is an >effective rate of around 9000 or so baud. I haven't tried it with a Mac since >I don't have one within cable length. > >ZLink, at least the version I looked at, patches around ProDOS to do its >interrupts and at least the version I tried left the machine in an undefined >state that required a power off/power on state change to recover from so a >program that followed the rules (i.e. TIC) could be used on the machine again. >Maybe this is similar to the problem AE Pro has where it intercepts system >interrupts before they get to the OS but doesn't reinstall interrupts when >finished. I don't think the problem is exactly like AE Pro since TIC can fix >what AE Pro messes up by restoring the page 3 vector. I have now done things the "legal" way by calling ALLOC_INTERRUPT. I am concerned though that the overhead of ProDOS will make 19,200 baud hard to achieve. If am were to tell MacKermit or Macterminal to xmit at 19,200, will it send data at 19,200 or does it have a lower effective rate (as you suggested in your original note is the case with //s)? I may be able to test high speeds with a nearby Mac. I have no idea why other comm programs couldn't recover from what Z-Link set up for itself. I did tinker with $3FE/$3FF, but I restored the original values at quitting time. I also poke around with the serial hardware directly, but I don't do anything out of the ordinary - just set the baud and so on. My IRQ handler is better than yooorrss :-) :-) I buffer 1k because the //c really lost when I was doing Ymodem xfers higher than or at 9600 baud. With the overhead of ProDOS now added, things may get worse though. We shall see. David Whitney, MIT '90 Still learning about my Apple //GS {out there}!harvard!think!whitney and all of its secrets. Any and all whitney@think.com technical info appreciated. DISCLAIMER: If they only knew what I was doing and saying here...
blume@netmbx.UUCP (Heiko Blume) (07/05/88)
i tested things at 19200 with a USR HST modem's help screen which comes at full speed. prgms with bigger buffers like z-link can keep up with buffering the input, but displaying is another story... with analog loop back test all prgms loose it sooner or later as the buffers overflow. buffers cant be big enough ! btw: if one slows down the cursor blink rate in z-link the prgm wont keep up with 19200 anymore.. -- Heiko Blume # DOMAIN: blume@netmbx.UUCP { BITNET: ( mixed } Seekorso 29 # BANG : ..!{backbone}!netmbx!blume D-1000 Berlin 22, West-Germany # Phone : (+49 30) 365 55 71 or ... 365 75 01 Telex : 183008 intro d # Fax : (+49 30) 882 50 65