Project

General

Profile

Bug #1239

CartesianProductList: case of just 1 entry in list

Added by John Abbott about 5 years ago. Updated about 5 years ago.

Status:
Closed
Priority:
High
Assignee:
Anna Maria Bigatti
Category:
bug
Target version:
CoCoA-5.3.0
Start date:
29 Jan 2019
Due date:
% Done:

100%

Estimated time:
0.99 h
Spent time:
1.00 h

Description

I believe CartesianProductList handles wrongly the case of a list with just 1 entry:

>>> CartesianProductList([0..1, 0..1, 0..1]);
[[0,  0,  0],  [0,  0,  1],  [0,  1,  0],  [0,  1,  1],  [1,  0,  0],  [1,  0,  1],  [1,  1,  0],  [1,  1,  1]]
>>> CartesianProductList([0..1, 0..1]);
[[0,  0],  [0,  1],  [1,  0],  [1,  1]]
>>> CartesianProductList([0..1]);
[0,  1]

Surely the last result ought to be [[0], [1]], right?

History

#1 Updated by John Abbott about 5 years ago

The bug would found by a student in Passau (Bernhard Andrashko).
His program misbehaved when computing CartesianProductList with a single "factor".

#2 Updated by Anna Maria Bigatti about 5 years ago

  • Assignee set to Anna Maria Bigatti
  • % Done changed from 0 to 10

hmmmmm, true.
I had been reasoning too mathematically....

I do not much like it though.
should instead return an error? (or maybe just print a warning?)
Do you remember how that was used?

Anyway, I have fixed it in my copy, now.

#3 Updated by John Abbott about 5 years ago

  • Status changed from New to In Progress
  • % Done changed from 10 to 20

I can try asking the student again for a copy of his program. However, the student was fairly good at programming, and I was mystified when he inserted code to handle a "special case" when I fully expected the general case to work (but it did not because of the bug in this issue).

Given my expectation, I think the case of CartesianProductList(L) where L is a list of a single list should work without warning (and certainly it should not give an error). While I understand the thinking behind "protecting a programmer from a careless mistake", I am not happy about a design which makes it more difficult to write a correct program (i.e. needing special handling for the case of a cartesian product of 1 factor). I also not that the function product works fine when given as argument a list containing a single entry.

#4 Updated by John Abbott about 5 years ago

I do note that CartesianProductList([]) returns [] without warning (or error, obviously).

I might be willing to be convinced that this case ought to product a warning/error...

#5 Updated by Anna Maria Bigatti about 5 years ago

  • % Done changed from 20 to 50

OK, convinced.
If he actually wrote a progam where he needed to handle the 1-case (and had to design his own workaround) then we fix it. Already done.

For the empty list I'm flexible. Now I think that the 0-case is OK and should give no nasty surprises:

/**/ CartesianProductList([[],[],[]]);
[]
/**/ CartesianProductList([]);
[]

#6 Updated by John Abbott about 5 years ago

  • Status changed from In Progress to Resolved
  • % Done changed from 50 to 80

Good, we're agreed. I've fixed it too -- we'll merge when I'm in Genoa.
I do still have a minor doubt about the empty product, but it's not that important.

#7 Updated by John Abbott about 5 years ago

  • Status changed from Resolved to Closed
  • % Done changed from 80 to 100
  • Estimated time set to 0.99 h

JAA notes that the product of zero factors should return a list containing 1 element which is a 0-tuple (namely, the list of the empty list). Code corrected.

Checked in. Closing.

Also available in: Atom PDF