Calling call//1 makes DCG nonfunctional? (+ "difference list" in DCGs vs. "difference list" for appending)

Help!
,
8

Can you give some code with examples as unit test to show the difference?

For me there are

  1. list - which are closed list and end with [].

  2. difference list using two separate variables - where the second variable is unbound until bound with a new difference list of two separate variables but then for the new difference list the second variable becomes the second unbound variable of the difference list bound to the second variable of the first difference list. I have a graphic of this but I don’t know where it went; might have meet the circular file. These are the form used with DCG and EDCG. With these I tend to think of the two variables holding pointers with all of the operations with these as just pointer manipulation which is what makes them so fast; you never have to copy strings.

  3. difference list where the first variable holds a complete list and the second variable holds a subset of the end of the complete list. These I don’t use other then with examples and have not found a use for them outside of examples. These might not even be a separate type of difference list but I just can not reason with them in my mind as I don’t have a unified view of difference list that incorporates this concept.

  4. difference list using a pair, e.g. A-B. - For me these are just a different syntax representation of 2 but keep the two variables together. In debugging EDGC at times I wished EDCG used these.

I am currently considering switching from 2 to 4 when not using DCG and EDCG based on this post. However since 2 is ingrained with DCG and EDCG will use that form when required.

EDIT

Found the graphic but because it was highlighted in color so it needs to be scanned in color to see correctly and that shoots the size of the file into the 10 of millions of bytes which for a storage to knowledge ratio is way to low. Someday I plan to do some of these using GraphViz but not for the foreseeable future.

EDIT

In Difference Lists by Frank Pfenning on page L11.5

Seasoned Prolog hackers will often break up an argument which is a difference list into two top-level arguments for efficiency reasons.

1 Like