Re: Remarks on SWI 7

Hi Jan and folks,

It is good that this community *does* discuss the issues that arise and we are all benefited in that Jan does hear and listen to the many voices here - and responds so very quickly. Thank you Jan - and all who contribute code. 

That being said - very much appreciatively - I haven't had a chance to check my existing body of code developed with a vers 5 instance of SWI with this new v7, but, I *would* be unhappy if the tens of thousands of lines of Prolog code I have that's invested in using SWI-Prolog needed a lot of rewriting or hand-walking to find things that are now broken as a result of the changes. As it happens, after quite a break from working on an expert system in SWI-Prolog, I am due to get back to it later this week hopefully, so, maybe I'll load up 7 on a non-critical system and see if everything still works and report back on that too as time permits.

At the very least, I hope that the older pre-7 instances might hang around a while while the issues with 7 - if issues they be - are worked through in terms of discussion and change as need be, and while I'm not promoting the idea - I prefer a well-maintained -traditional switch approach albeit I understand that it's extra work - it's possible that we would end up seeing a fork if things begin to diverge too much from historical instances of Prolog. That would probably not be good.

So, for $0.02, if there is a way ongoing Jan to make sure that what has worked continues to, all the while adding good new things which are appreciated - map looks very useful to me for example - it might be wise to observe that there is a variety of opinion being voiced about this and consider how to tackle that a tad further.

With thanks for all your effort in any event!

Kind regards

Derek Jones. 



On Nov 26, 2013, at 1:10 PM, Ulrich Neumerkel <[email protected]> wrote:

> Some remarks on the appearance of SWI 7.
> 
> In the past there have been discussions on language features on this
> list and elsewhere.  In this context I maintained the position that
> many of the innovative features in SWI can be crafted such that they
> conform to ISO standard ISO/IEC 13211-1:1995 as an extension, that is,
> according to subclause 5.5 Extensions.  In particular, the very
> interesting ideas to better support text could be realized within ISO.
> 
> The ISO standard was never meant as the last word on Prolog but as a
> solid foundation for existing investments and further developments.
> Recent corrigenda (Cor.2:2012), removed the very last obstacles for
> DCG and CHR.  That is, both DCGs and CHRs can now be defined without
> redefining token syntax. And there are cleaner ways to handle
> variables which benefits constraints.
> 
> It is quite feasible to conform to ISO.  In fact, the very first
> conforming systems appeared with the publication of the standard in
> 1995.  And these systems are even today more conforming than SWI ever
> was.  Also the recent corrigenda have been taken into account widely.
> SICStus 4.3, GNU 1.4.4 come to mind who now conform a 100% when
> reading syntax.  Id est: I **know** 0 (zero) differences.  Same for
> other areas, I have not tested everything. Don't hesitate to
> report any differences if you know them.
> 
> http://www.complang.tuwien.ac.at/ulrich/iso-prolog/conformity_assessment
> 
> SWI 7 now gave up most of Prolog syntax and semantics.  And by that I
> mean the common subset between ISO Prolog and predecessors all the way
> down to DECsystem 10 Prolog of 1978 (that's the first manual's date).
> There **is** a lot in common in all these systems.  Maybe heated
> discussions have blurred that aspect.  A typical Prolog class/textbook
> will never go to such detail as to identify a difference.  Helas, SWI
> never supported that common subset in full.  And it seems that SWI 7
> shifts into non-relational properties as defaultiness.
> 
> Now, SWI 7 jettisoned that common subset.  Maybe that makes sense,
> maybe not.  I know not.  It is certainly a possible research path.
> What I do know for sure is that this system is no longer Prolog.  In
> the same sense that, say, Clojure is no longer Lisp, although it is
> considered "a Lisp".  It would be fair to reflect that in the name of
> SWI 7.
> 
> From a personal viewpoint I am certainly disappointed after spending
> some years in improving SWI in the hope that I will one day be able to
> use it as a stable platform.
> 
> All that said, let us not forget the breakthroughs and victories of
> SWI in the last years: occurs_check, the ingenious algorithm for
> unify_with_occurs_check/2 with infinite trees which still awaits a
> formal treatment, library(clpfd), library(pio) - regrettably never
> completed, setup_call_cleanup/3 - and some more.
> _______________________________________________
> SWI-Prolog mailing list
> [email protected]
> https://lists.iai.uni-bonn.de/mailman/listinfo.cgi/swi-prolog

-------------- next part --------------
HTML attachment scrubbed and removed