Python list vs google group

[Richard Damon]

On 6/15/18 11:07 PM, Jim Lee wrote:
> 
> 
> On 06/15/2018 07:08 PM, Richard Damon wrote:
>> On 6/15/18 9:00 PM, Jim Lee wrote:
>>>
>>> On 06/15/2018 05:00 PM, Chris Angelico wrote:
>>>> On Sat, Jun 16, 2018 at 4:52 AM, Rob Gaddi
>>>> <rgaddi at highlandtechnology.invalid> wrote:
>>>>> On 06/15/2018 11:44 AM, Larry Martell wrote:
>>>>>> My favorite acronym of all time is TWAIN
>>>>>>
>>>>> Really?  I always thought it didn't scan.
>>>>>
>>>> Having spent way WAY too many hours trying to turn documents into
>>>> images (and text), I very much appreciate that laugh.
>>>>
>>>> ChrisA
>>> I once had a Mustek color scanner that came with a TWAIN driver.  If
>>> the room temperature was above 80 degrees F, it would scan in color -
>>> otherwise, only black & white.  I was *sure* it was a hardware
>>> problem, but then someone released a native Linux driver for the
>>> scanner.  When I moved the scanner to my Linux box, it worked fine
>>> regardless of temperature.
>>>
>>> -Jim
>> There actually may still have been a hardware issue, likely something
>> marginal in the timing on the cable. (Timing changing with temperature).
>> It would take a detailed look, and a fine reading of specs, to see if
>> the Windows Driver was to spec, and the hardware is marginal (but the
>> Linux driver didn't push the unit to full speed and got around the
>> issue), of if the Windows driver broke some specification but still sort
>> of worked, especially if things were warm, while the Linux driver did it
>> right.
>>
> You are exactly right.  It was so long ago that I forgot some of the
> details, but it boiled down to the TWAIN driver pushing the SCSI bus out
> of spec.   I remember looking at the bus on a scope, measuring pulse
> widths, and playing with terminator values to try to optimize rise
> times.  I don't believe it used the Windows SCSI driver at all, but
> instead drove the scanner directly (I could be remembering wrong).
> 
> -Jim
> 

That sounds like it would probably be classified as a software issue
then (or possibly documentation). It could be hardware if the Windows
SCSI card didn't support something it was expected to or perhaps
indicated that it did, or didn't negotiate correctly.

Ultimately, often the difference between a hardware error and a software
error is what the documentation says, I have seen more than once a
hardware document saying something like Feature A was intended to work
this way but the hardware doesn't work right to implement it, so the
software needs to do XYZ as a work around. So now, if the software
doesn't do XYZ it is a software error, all due to a hardware design
issue that was just redefined.