On 01/21/2013 07:06 AM, Ferrous Cranus wrote:
<snip>
Seriously, you're asking for something that's beyond the power of
humans or computers. You want to identify that something's the same
file, without tracking the change or having any identifiable tag.
That's a fundamentally impossible task.
No, it is difficult but not impossible.
It just cannot be done by tagging the file by:
1. filename
2. filepath
3. hash (math algorithm producing a string based on the file's contents)
We need another way to identify the file WITHOUT using the above attributes.
Repeating the same impossible scenario won't solve it. You need to find
some other way to recognize the file. If you can't count on either
name, location, or content, you can't do it.
Try solving the problem by hand. If you examine the files, and a
particular one has both changed names and content, how are you going to
decide that it's the "same" one? Define "same" in a way that you could
do it by hand, and you're halfway towards a programming solution.
Maybe it'd be obvious from an analogy. Suppose you're HR for a company
with 100 employees, and a strange policy of putting paychecks under the
wipers of the employees' windshields. All the employee cars are kept
totally clean of personal belongings, with no registration or license
plates. The lot has no reserved parking places, so every car has a
random location.
For a while, you just memorize the make/model/color of each car, and
everything's fine. But one day several of the employees buy new cars.
How do you then associate each car with each employee?
I've got it - you require each one to keep a numbered parking sticker,
and they move the sticker when they get a new car.
Or, you give everyone a marked, reserved parking place.
Or you require each employee to report any car exchanges to you, so you
can update your records.
If you can solve this one, you can probably solve the other one. Until
then, we have no spec.
--
DaveA
--
http://mail.python.org/mailman/listinfo/python-list
