Shawn Wright said:
Actually, when it comes to scanners, NT is not all that different from
win9x,
as it uses essentially the same Twain driver interface. (one of the big
weaknesses in scanner support, imho). I have yet to find a SCSI scanner
which does not work in NT (including about 10 different models from
Epson,
HP, UMAX, Acer, Microtek), since the Twain drivers are usually the same
ones used in win9x. In many cases, the SCSI card drivers are also common
between win9x and NT.
----------------------------------------------------
Twain drivers are an attempt to sort out the cross-platform image device
support problem but the Twain driver itself is only part of the
solution. A Twain driver provides a standard software interface from
the application program to a virtual piece of hardware. It attempts to
make all of the disparate hardware devices speak the same language on
the application software side. Underneath the virtual hardware,
however, is a piece of real hardware which must be controlled by a Twain
compliant hardware device driver... ie; the stuff that actually controls
the hardware. This could be, for example, a specific SCSI driver that
speaks Twain on one side and translates it to specific SCSI hardware
control signals on the other. However, this SCSI driver has to be
specific to Win/9x or Win NT since the way device drivers interoperate
with these operating systems is totally different.
Installation procedures might give the appearance of using the same
hardware device drivers between Win9x and Win NT but the actual driver
software is as different as night and day.
If Olympus uses Twain and SCSI it reduces the cost of supporting both
operating systems but doesn't eliminate extra cost. If Microsoft
ultimately produces a common device driver model so that only one driver
is required the costs are reduced even further but still not
eliminated. The next hurdle is testing which is also an expensive
process. Even though the systems might be supposed identical, in
reality, they often aren't because of minor glitches in the
implementation.
On the importance of testing, need I remind anyone of the problem with
the Hubble telescope? A perfectly ground and polished lens but ground
ever so slightly to the wrong specification. And the reason it wasn't
tested was it saved approx. 1 million dollars. Testing is an expensive
process even when the code for two environments is identical. For small
markets it might be the only thing standing between statements of
support or non-support.
Chuck Norcutt
Woburn, Massachusetts, USA
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|