Project

General

Profile

Feature #674

Ring constructor: allow empty range for indices?

Added by John Abbott about 9 years ago. Updated over 3 years ago.

Status:
Rejected
Priority:
Normal
Assignee:
John Abbott
Category:
enhancing/improving
Target version:
CoCoA-5.4.0
Start date:
13 Mar 2015
Due date:
% Done:

100%

Estimated time:
1.21 h
Spent time:
1.20 h

Description

I am writing a CoCoA-5 program which may need some "dummy" variables. My code looks like this:

NumDummies := ...;
P ::= QQ[x[1..9],dummy[1..NumDummies]];

This code works fine if there is at least 1 dummy variable, but fails if there are none. The dot-dot operator complains that the end of the range is smaller than the start.

Should we allow empty ranges?

History

#1 Updated by John Abbott about 9 years ago

There would be no technical problem allowing a range such as 1..0; perhaps the main question is whether the gain in extra flexibility is "greater" than the reduction in safety (protecting a user from carelessly inputting a daft range).

If we do allow empty ranges, must they be of the form n..(n-1) or would we also allow "very empty" ranges such as 2..0 or in general n..(n-k) for some k > 1?

#2 Updated by Anna Maria Bigatti about 9 years ago

John Abbott wrote:

There would be no technical problem allowing a range such as 1..0; perhaps the main question is whether the gain in extra flexibility is "greater" than the reduction in safety (protecting a user from carelessly inputting a daft range).

I think we should not allow empty ranges.
I find it more likely that a user writes x[4..1] thinking to get [x[4],x[3],x[2],x[1]]

#3 Updated by John Abbott about 9 years ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 10

I understand your point. Actually, I suppose the correct solution is to use anonymous indets for the dummies, but we do not have those in CoCoA-5.

So how do I make my code work? I could create at least 1 dummy variable (even if I will never use it), or I could put in an if statement and use two different commands to build the ring I want to work in. Neither solution appeals to me much.

You would not be willing to compromise:
  • n..(n-1) is OK and gives the empty range
  • n..(n-k) for k > 1 gives an error

#4 Updated by Anna Maria Bigatti about 9 years ago

John Abbott wrote:

I understand your point.

Actually, I suppose the correct solution is to use anonymous indets for the dummies, but we do not have those in CoCoA-5.

That's another problem, but we decided it is not worth having anonymous symbols in cocoa-5
(and I think it was a sensible decision)

So how do I make my code work? I could create at least 1 dummy variable (even if I will never use use), or I could put in a if statement and use two different commands to build the ring I want to work in. Neither solution appeals to me much.

I would go for the second. The code might in principle be different if you have no dummies.

You would not be willing to compromise:
  • n..(n-1) is OK and gives the empty range

I'd rather not: x[2..1] in my head is closer to [x_2, x_1] than to []

  • n..(n0k) for k > 1 gives an error

?

#5 Updated by John Abbott over 3 years ago

  • Assignee set to John Abbott
  • Target version changed from CoCoA-5.?.? to CoCoA-5.4.0
  • % Done changed from 10 to 30

This was last considered 5 years ago. The problem has not arisen in this time (in my experience), so I think that it is a rare occurrence.
The logical conclusion would be that there is no real demand for empty ranges -- in those rare cases where they might be needed, simple (if tedious) workarounds exist.

As a consequence I suggest rejecting this proposal. Anna?

#6 Updated by Anna Maria Bigatti over 3 years ago

  • Status changed from In Progress to Rejected
  • % Done changed from 30 to 100
  • Estimated time set to 1.21 h

Also available in: Atom PDF