On 28/08/11 5:05 PM, Dave McGuire wrote:
  On 08/28/2011 04:44 PM, Toby Thain wrote:
   >      And those things are the ones I'm talking
about.  We will
> never lose
> the spec for the JFIF file format because the whole world uses it.  We
> will also never lose (assuming we have it now) the PERQ image file
> format because there are people like us around.
 I don;t beleive any of us are immortal, and nor are the people who get
 our collections when we pass away. It's always possible that at some
 point in the not-too-distant future one of said nth-generation
 inheritors
 says 'This pinoe of paper is worthless', and then some file format (or
 whateer) is lost forever. 
    Good heavens.  Do you really think it takes a bunch of hobbyists
 maintaining obscure information to preserve, say, the spec for GIF,
 JFIF, and PNG? 
 
 These are terrible examples and illuminate very little. Why do people
 insist on using them? 
 
   As examples in this discussion?  Because we were talking about common
 image formats, and someone asserted that all image formats are evil
 because the specs for everything will be lost. 
I think we are all agreed that JPG, PNG, GIF will not be lost. Now let's
talk about the *interesting* cases...
   As such, with the vast, VAST majority of images in the world today
 being in one of those formats, it is actually an excellent example.
     In the
case of file formats that actually get used NOW,
 hell, those specs are printed in dozens of books by several different
 publishers in many different languages, and sold by the thousands.  And
 that's not even counting digital archives. 
 That's simply not true. Many formats are proprietary. 
 
   Formats that people actually USE?  No.  Sure, there ARE some
 proprietary formats out there,  
More than a few.
  and I'm sure there are some poor saps who
 use them from time to time.  Your insistence that these obscure cases
 are somehow the norm is amusing.
   For the formats that actually get used, unless we undergo some sort of
 a friggin' holocaust or something, said documentation will always be
 available.  And even if we DO have some sort of a holocaust, I'm sure
 somewhere there will be at least ONE remaining copy of one of the
 hundreds of thousands of books printed on the subject.
   Even vertical-market imaging formats used in the world of astronomy,
 physics, and medicine are (mostly) standardized and documented these
 days.  Stop with the doom-saying.
  Even extensions
 and codecs used with extensible "open" formats are frequently
 proprietary (e.g. TIFF). 
   Yes.  And the 10% of the world that uses TIFF (as nice a format as it
 is) has maybe 10% of THAT using proprietary extensions.  People that use
 them should probably move to a documented format.
  One *extremely* common format (PSD) is very
poorly documented. The vast
 majority (>  90%) of what PSD is supposed to represent is not documented
 at all. Yet data that cost billions of dollars to create rests in it. 
   Gee, I open PSD files in GIMP all day long, works just fine here.
 Maybe there are some weird features that won't be properly decoded, but
 I can SEE THE DAMN IMAGE just fine, and isn't that the point, in the
 context in which we're discussing this? 
There are plenty of PSDs that Gimp cannot open properly. And even those
that Gimp *can* open, only a small subset of features are supported.
   But even ignoring that...your supposed all-destructive holocaust is
 going to destroy EVERY REMAINING COPY of Adobe Photoshop, along with
 EVERY COMPUTER capable of running it? 
Yes, that's very close to what inevitably occurs, once you put thought
into the problem. You are dealing with obsolescence on several levels
simultaneously, among them:
 - media
 - file format
 - application
 - operating system
 - machine architecture
These all go obsolete on rather short cycles. It is not necessary for
the "last copy" to be destroyed. It is only necessary for "a copy I can
easily get hold of, along with a working machine of that architecture I
can easily get hold of, along with a copy of its system software, ..."
etc, etc. And soon enough you begin to need knowledge that you either
forgot or never actually had. Like a serial number. Or remote server
activation. <GAME OVER>
  TL;DR: Just because something is ubiquitously
used today does NOT mean
 the knowledge to decode it is known outside tiny, doomed cabals. 
   This is just ludicrous.  Some peoples' obsession with pointing out
 obscure corner cases and insisting that they are somehow the norm is
 entertaining at first, but very quickly becomes old. 
 
PSD isn't an obscure corner case - it's (for better or worse) an
industry standard. And we could cite plenty of others, that are not
publicly documented.
--Toby
               -Dave