The New Sudoku Players' Forum

[denis_berthier]

.
Today's update follows a bug fix in CLIPS 6.4 (at my request) in the CLIPS output function to the standard output.

Changes to CSP-Rules are only about "progression bars" in the few functions dealing with files of puzzles: after the fix, they display a "." after each line is dealt with, as used to be the case with CLIPS 6.3.

If you're still using 6.3, this update brings no change.

Note 1: Only the CLIPS updates that have an impact on CSP-Rules are integrated into the CSP-Rules copy. It happens very rarely, as CSP-Rules uses only a part of CLIPS.
Note 2: If you're using CLIPS 6.4, in addition to the previous two "errors" printed when loading CSP-Rules, a third one will now appear, for function "flush". As before, it can be ignored.

[Edit]: If you've already downloaded the update before seeing this edit, you'll have to do it again: there was a bug in a CLIPS 6.4 default setting, which entailed very slow computation times.
.

[denis_berthier]

You don't update CSP-Rules every time CLIPS is updated. Which version of CLIPS should I use?

SudoRules, as all of CSP-Rules, is totally written in the CLIPS language (itself totally written in C for efficiency). CSP-Rules uses only part of the CLIPS system - namely only the inference engine (to solve puzzles) and part of the Lisp-like functional language (to input commands into CSP-Rules, to process them, to do stats and to output results).
I don't update the versions of CLIPS included in CSP-Rules when this would have no impact on CSP-Rules - i.e. most of the time in the past few years.
There are basically 2 versions of CLIPS compatible with CSP-Rules: 6.3 and 6.4, both delivered in the CLIPS folder of CSP-Rules. You can use any of them; it will have no impact on speed or whatever. You can also download the latest release of CLIPS 6.3 or 6.4 from their website and compile it yourself (with command "make" in the "core" folder); it will not change anything for CSP-Rules.

There's also a new 7.0 version of CLIPS, still under development (and currently not included in CSP-Rules). I've tried it and it works as well as the 6.3 and 6.4 versions. Except that the inference engine part is slower (by about 10% during solving puzzles, according to quick calculations with SudoRules). This is understandable, because it adds possibilities to the underlying RETE network. As the new functionalities are not relevant to CSP-Rules, I don't plan to add this version in the near future.
.

[denis_berthier]

.
I just fixed a bug in the GitHub repository. It could have prevented the correct ARM version of gsf's software to be used on an ARM Mac. (As x86 code will not be supported by MacOS in the future, this could have become a real problem).
.