Re: Is ProDOS supported in OSX?

[pausch@saaf.se (Paul Schlyter)]

In article <20030625193637.16978.00001693@mb-m06.aol.com>,
Michael J. Mahon <mjmahon@aol.com> wrote:
> Paul Schlyter replied:
> 
>>In article <20030625041241.09942.00000955@mb-m26.aol.com>,
>>Michael J. Mahon <mjmahon@aol.com> wrote:
.........................
>>> Of course, if some new compression technology comes along
>>> that buys another factor of five or more, then someone may be
>>> willing to risk de- and re-compression (generally to be avoided).
>> 
>>Scientific data should preferably NOT be stored in some lossy
>>compression format, like JPEG ......
> 
> Agreed.  Not my example, but one of yours.  ;-)
> 
> But the ideal of lossless compression is probably not the best
> tradeoff for data sent back over a data-rate-limited link from a
> great distance.  There the choices are: fewer pictures with
> lossless compression or several times as many pictures
> with lossy compression.
> 
> If there are several overlapping frames of the same features
> compressed with a lossy format, it is not too difficult to isolate
> (or at least bound) the errors due to the lossy format by comparing
> different frames containing the same feature.  And, of course, the
> degree of "loss" in most lossy formats can be varied, so that one
> frame in ten, say, is transmitted with little or no loss for reference.
> 
> If I were imaging Pluto, I would prefer, say, six pictures per unit
> of transmission time with five of them using modest lossy
> compression to two pictures with lossless compression.
 
JPL had precisely that problem with Galileo: when its high-gain
antenna failed, the JPL engineers suddenly had to face the fact that
they only had a fraction of the bandwith available than what they'd
expected.  They attacked this problem in several ways -- adding data
compression was one way.  But they never went as far as introducing
lossy data compression -- the integrity of the data is very essential
in science.
 
So even if you'd receive, say, three times as many images by using
lossy data compression, those images would be full of artefacts,
i.e. false details due to the data compression.  The press, who
merely wants pretty pictures, wouldn't mind of course.  But the
scientists, who sometimes examine the pictures pixel by pixel,
would frown strongly at this!
 
> Another, perhaps optimal, strategy would be to transmit all
> frames promptly with lossy compression while retaining them
> in storage, and then, after all frames were transmitted,
> re-transmit them lossless as time and transmission bandwidth
> permits.
 
On Galileo there was another problem: the onboard storage was
limited....  only control signals were transmitted promptly,
definitely no images since they required far too much bandwitdh.
The rest was recorded onboard -- and during later, more "quiet"
periods of that space probe, the recorded data was transmitted
to Earth, slowly.
 
> One could even use the lossy frames as an "index"
> to request particularly interesting frames to be sent sooner
> in lossless, maximum resolution mode.
 
It's better to use low-resolution images as index images: going
from lossless to lossy compression will compress the data by
an additional factor of perhaps 4 (depending on HOW lossy the
compression is of course).  Reducing the resolution by a factor
of 2 in both direction will yield the same result.
 
>>> And if we're talking about the _really_ long term, say thousands
>>> of years, then I feel safe in saying that the programs necessary
>>> to decode data will be recorded as a brief preface to the data,
>>> in an abstract but machine-interpretable form.  This is in many
>>> ways the ultimate kind of meta-data--an operational description
>>> of the data that follows.  (Or, an abstract object-oriented data
>>> representation, if you will.)
>> 
>>If we're talking about thousands of years, we probably need to be
>>concerned about the survival of our culture.  Go a few thousand years
>>back and you might end up in pharaonic Egypt; their language was
>>lost for many centuries, and the writings they left behind remained
>>undecipherable until fairly recently.
>> 
>>Unspecified digital data will be even harder to decipher than
>>egyptian hieroglyphs.  This includes opcodes for some abstract
>>machine.
> 
> If the culture is interrupted suddenly and for a considerable
> time, then this is perhaps the ultimate risk.  (If the interruption
> is not sudden, then there are several extraordinary things
> which could be done to improve the odds of survival of a
> "bridge" to the past--multiple "mega" time capsules come
> to mind, perhaps some on the surface of the Moon.)
> 
> First, we must ask how much stake we have in a future
> culture, discontinuous with our own.  Perhaps they must
> (and should) learn for themselves and make their own
> mistakes.  ;-)
 
It's also a matter of how interested we are in the far future.
Consider for instance environmental issues -- if they're not
likely to do much harm to our generation or the next generation,
most people tend to dismiss them as unimportant, even if the
really long-term damage culd be severe.
 
> If we cannot assume cultural continuity, then the problem
> of communication is much like the problem of communicating
> with an extraterrestrial intelligence.  We must assume only
> the minimum of shared knowledge--like mathematics, physics,
> and astronomy--and then build out from there.  Data format
> description is probably the least of our worries.  I would expect
> that finding a way to describe what the data represents is
> harder than describing how the data is represented.
 
Communicating with future cultures are easier on one way though: we
can assume they'll get access to physical objects we leave behind.
 
-- 
----------------------------------------------------------------
Paul Schlyter,  Grev Turegatan 40,  SE-114 38 Stockholm,  SWEDEN
e-mail:  pausch at stockholm dot bostream dot se
WWW:     http://www.stjarnhimlen.se/
         http://home.tiscali.se/pausch/