Howard Brown escribió: > GM Jose, > > There is one point in your post I would like to bring up. Where you > say: > > >"Multitasking cannot handle tight ARQ timing windows. > > >It is a pity that noone has come forward (as far as I know) a real > >time OS (RTL, for instance) with a proposal usable on an old PC as > >packet engine with a sound card as modem. The problem is not the PC > >itself, but the prevalent OS's." > > Why not challenge the need for the tight timing windows? This > creates more wear and tear and the radio but what real benefit does > it provide? I believe that the NBEMS package and the RFSM8000 > package prove that you can have effective ARQ without the fast > switching.
Examples of tight timing and succesful implementation are P-II and P-III, not only AMTOR, with its hair raising clicking.... I never tried SCAMP, which seemingly did well running under Windows. Actually, I have entrenched on my mind what gave me good results, but certainly, that may not be the only way for success, I have to admit. Examples of the success of tight timing are WSPR and WSJT, and certainly, knowing WHEN to expect input is a bonus. None of them generate such a high wear and tear. WHEN may be absolute, like in WSJT, related to the UTC time scale, or relative, in some time measure after the last received packet. I have had little luck with RFSM8000, because of my perennial lack of time in recent times and the little user mass it has generated. > I would love to see Linux used more and not need to deal with Windows but still, the quick switching seems to be of little value. It was a way, which certainly, has been proven not to be the only one. You cannot entirely disegard trends, that may vary in time. I was the sysop of three BBS's at a time, on MSDOS (even with Desqview "multitasking"), Linux and Windows (the last one was a quick hack and an exercise in lazyness) and I liked BBS's. Forwarding over radio links, with all the freedoms it provides, and some associated extra responsibilities too. But I fail to see the comeback of the BBS's. Winlink may be useful, but it hardly substiututes the "packet" network the way it operated in the 90's. I have very good memories of the Linux based systems I ran (node, FBB, JNOS, DXNET), both were highly resilient and almost bulletproof. Uptime was usually more than 30 days, and the PC's ran without battery backups. I always used hardware TNC's, Kantronics and SCS. Pactor allowed 1 MB per day of forwarding on HF easily. Packet (300 baud) hardly ever exceeded 100 kb per day. 73, Jose, CO2JA