Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!motcsd!hpda!athertn!paul From: paul@athertn.Atherton.COM (Paul Sander) Newsgroups: comp.sys.apple Subject: Re: ZMODEM/EASYLINK Summary: Takes longer than 4ms Message-ID: <9815@athertn.Atherton.COM> Date: 2 Aug 89 18:37:25 GMT References: <8906180052.AA20316@obsolete.UUCP> <10417@smoke.BRL.MIL> Organization: Atherton Technology, Sunnyvale, CA Lines: 24 In article <10417@smoke.BRL.MIL>, gwyn@smoke.BRL.MIL (Doug Gwyn) writes: > This isn't an inherently insoluble problem, because virtually all UARTs, > ACIAs, or whatever in use today are at least double-buffered, giving the > operating system one character assembly time to deal with inputting the > buffered character. At 2400bps, that's 1/240 seconds, i.e. more than 4 > milliseconds. If interrupts are disabled only for shorter duration than > that, no received characters need be lost. > > I don't know how long ProDOS or GS/OS disables interrupts during I/O for > the various kinds of supported disk drives, but all of them should be > able to transfer a sector in less than 4 msec. True, they _should_ be able to transfer a single sector in 4ms. The problem is that most communication programs (all of the ones I've ever used, and all of the ones that people I know have used) tend to buffer up the data and write many sectors at once. Worse, for those who transfer to floppy, the drive has a 1-second spin-up time. I don't know about anyone else, but I wouldn't count on interrupts being enabled during that time, at least not when using a Disk ][. -- Paul Sander (408) 734-9822 | If a machine is powerful enough paul@Atherton.COM | to have a DWIM button, why bother {decwrl,sun,pyramid}!athertn!paul | with the button? -- Eric Black