What do I look like, a worker of miracles?

[vakkotaur]

I like to believe I'm a fairly good programmer. I don't claim to be Wile E. Coyote, programming super-genius. But evidently some folks at work figure I can re-program external reality. (If I could, do you think I'd have left it in the condition it's now in?) This morning was "How is the ____ program coming along?" in regard to a problem that showed up yesterday afternoon during what should have been final testing.

After some digging and some more testing, I was able to demonstrate that the program actually did do exactly what was asked. Actually, two programs on two different pieces of hardware that had to talk to each other. Connected by cable, they worked just fine. Connected by a radio link, a necessity for this particular project, things got shaky. I have exactly zero control over the radios. But somehow I'm supposed to fix them in software on other hardware. I don't think so.

It turns out that not only were the radio-modems never properly configured, the guy who was setting them up didn't have the configuration tool for it. And that should have been a clue - we stopped using that brand of radio-modem a couple years ago. Gee, maybe there's a reason we switched brands? That was pointed this morning. Now, this afternoon, it finally dawns on the guy with radio-modems that maybe he should use the radio-modems that work.

Comments

[altivo.livejournal.com]

Well, the programming for packet radio in the amateur bands does indeed compensate for bad modems and lousy connections. It does so by using checksums and CRCs and requesting retransmission of every defective packet again and again until it comes through intact. Not terribly efficient for a commercial application, I'd have to agree. On the other hand, it seemed pretty amazing when first introduced because it worked under some really substandard conditions, albeit slowly.

I wonder if commercial services have picked up yet on the newest digital modulation methods in use by hams? Rotated phase shifts and such, the results are really amazing. The modems can read a signal so weak you can't hear it with your ears.

[vakkotaur.livejournal.com]

These are frequency hoppers. If they had been properly configured I expect there would not have been a problem. Since things are supposed to be in near real time, each time I send, the radio should send. Instead, the transmitter seems to have a buffer it needs to fill - or will send at predetermined intervals if the buffer isn't full enough. The amount the buffer must be filled in order to trip a transmit as well as the transmit-anyway interval can be set by configuration. Neither of these values was set to something that would work in this application. Had the configuration tool been available, useful values could be set.

I did wonder about handshaking, at least until I saw the receiver output on a terminal. It would be nothing, then a burst of several of my 'packets' all at once. I have to remind folks now and then that digital isn't necessarily more reliable than analog, but it can be far more stubborn.

I don't know if any of the commercial stuff has adopted stuff like BPSK, which I first encountered with the LOWfers some time ago. They were getting impressive results then. Now they've gone to some interesting receiver-side tricks and some rather simple though extremely slow keying. Much of it comes down to being able to have a patient machine listen rather than an impatient person.

[altivo.livejournal.com]

BPSK is amazing all right. It proves the possibility of communicating with very little in the way of RF equipment, relying just entirely on computers to do the signal processing. PSK31 is a more sophisticated example of the same principle.

Probably both seem too slow to anyone designing commercial applicatons, but I suspect commercial designers are missing bets. PSK31 handles pretty decent baud rates for transferring text or data, though it isn't up to broadband internet requirements of course. BPSK is really slow, but still might suit for long distance low power applications, like robotic spacecraft sending observation telemetry.

[kiwihunter8.livejournal.com]

"What do I look like, a worker of miracles?"

To them, you probably do, because you can do something very complex that they can't.

I get the same thing at my job, although programming is a lot more complex than what I do (graphics).

[altivo.livejournal.com]

Yeah. Amazingly enough, being able to code HTML without some bloated generator program also impresses the heck out of people, as if there was really anything to it.

When I worked as a system programmer, though, I found that most people couldn't even understand what I did, so they weren't that impressed. ;p

[jmaynard.livejournal.com]

What kind of a systems programmer?

[altivo.livejournal.com]

IBM OS2/MVS and later DEC VMS and RSX. Started out maintaining custom language compilers, moved into databases, Intercomm and CICS support, then DASD management. When they added VAXen and PDP/11s, I volunteered to learn to tune and manage them. At the next employer I began to learn UNIX and also did quality control engineering for CICS-based applications on both DOS/VSE and MVS.

I don't miss any of that, but it's an interesting background to have. I'm still amazed to have more computing power on my desktop than some of those mainframes had back in the 80s.

[jmaynard.livejournal.com]

Merciful $DEITY. I thought I was the only one who remembered Intercomm. I had to do that for a few years in the early to mid-80s. I also did MVS, CICS, and VTAM for many years.

Have you heard of Hercules? With it, and OSes from the CBT Tape Project, you can relive those thrilling days of yesteryear on your very own desktop.

[altivo.livejournal.com]

OMG! I had no idea. I assume this was done for some practical reason and not just for nostalgia's sake? I mean, I've seen the CP/M and TRS-80 and Amiga emulations. Come to think of it, there used to be a fairly functional VAX/VMS implementation too. But this definitely takes the prize. :)

[jmaynard.livejournal.com]

Lots of practical reasons...in particular, a lot of Linux/390 and z/Linux development is done on it. The nostalgia factor is certainly preent, though. The origiinal author did it at least in part so he could run MVT again.

[vakkotaur.livejournal.com]

What continues to surprise me is how often the simple and the complex seem to get reversed. Many is time I've heard the dreaded phrase, "All you have to do is..." which almost always is followed by something messy and complicated but someone else has the impression that it's something trivial that I can knock off in a couple minutes. Yet other times I am asked, "Can we...." which is often followed by something that really is a trivial change, but the other person isn't sure it's even possible.

[jmaynard.livejournal.com]

Indeed. When I was working on a payroll program for my mother's employer, I needed to distinguish the third biweekly check issued in a month from the first two for benefits withholding purposes. Mom thought that would be hard. I pointed out that it was quite simple: if it was issued on the 29th through the 31st of the month, it was the third check; if not, it wasn't. I had to get out a calendar and show her before she got it.