[AppauledAtAppeasementConservat]

"I have a hard time believing that the file sizes are that small."

I can't speak for JPL but back in the DOS days you could do an aweful lot with small files. Adding graphical interfaces and fancy stuff on cumputers takes up alot of space and that most likely would be useless on a robotic rover.

[Don Joe]

I can't speak for JPL but back in the DOS days you could do an aweful lot with small files. Adding graphical interfaces and fancy stuff on cumputers takes up alot of space

I wrote a DOS shell -- GUI, but character-mode-based -- that fit on two 360kb floppies (or one 720K floppy) for distribution.

It included a DOS menuing system, a GUI configuration program (for creating your DOS app menu entries), a graphical file manager (including a full licenced copy of PKZIP/UNZIP, with a GUI frontend I wrote), and an "in-program" cooperative multitasker (for maintaining the on-screen clock display and animated background).

It also included a Windows 3.x component that I wrote, to allow you to run Windows apps from DOS. My shell would launch Windows, run the selected application, and then sit silently in the background, monitoring your app. When you closed your app, my code would terminate Windows, and put you back in the DOS shell. (It also had options to stay in Windows when your app ended, or, query you as to whether to remain in Windows or return to DOS.) I added that because at that time, many people didn't have a lot of Windows apps, and only went into Windows for the express purpose of running a single program, and they'd go back to DOS when they were done.

Yes, you could do a lot in a little bit of space back then. (Both disk space, and RAM. My shell worked on computers with 640 KB RAM.)

I gave up on my shell when Windows 95 came out and I could see the handwriting on the wall. I'd had a distributor all lined up, too, hot to start publishing it, too. Timing is everything, argh.

[ambrose]

The old Apollo spaceships were run off of 16k computers!

[biggerten]

The rovers operating system (OS) is VxWorks from Wind River Systems. No GUI, and I'm sure very efficient at file storage.

Sensor data is most likely from 12 bit A/D converters, and they most likely pack two 12 bit readings into 3 bytes of storage to minimize file size. Critical here is data sampling rates - how many samples per second? And they can compress by various methods, such as run-length encoding, or most likely more sophisticated compression schemes. Data packing is a very intensive area of research, both for file size and 'squirting' data through during transmission where bandwidth and data transmission windows are limited. At the highest data rates, the rovers communicate at what - 110k baud? 11K bytes per second? That would take 48 hours for a complete flash image to be transmitted without compression - but at 110K baud, compression has already happened. And how many hours per day can they transmit? Certainly less than 12, and I bet a lot less than 12 hours.

The pictures are a problem (huge amounts of information - 8 or more bits per pixel? x3? ((three colors, don't forget!))) so 24 bytes per pixel on a color image, and I note JPL talks about loss-less images - the more compression, the greater loss of resolution (think MPEG, JPEG, MP3 and the like).

These are interesting problems. NASA is not capable of magic, but do have a ton of very smart people on staff. Smarter than those simpletons who snorted at Spirit's problems and knew, of course, if only they had been in charge nothing like the sort of what when wrong could have happened (obviously they have never been involved in a serious development effort).