The code fails to run due to the discontiguous predicate definition lines

[gabouelseoud]

I describe the problem in this thread and the colleague who replied also failed to get the expected output.
mthom/scryer-prolog#1014 (comment)
I hope you can help us

[flexoron]

At the home-page there are 3 examples where you can try and go:
https://www.metalevel.at/prolog/timetabling/

(Cut/Paste) Example 1 and 2 do work(but no timetable.txt has been written)

$ ../scryer-prolog/target/release/scryer-prolog simsttab.pl example1.pl OK!
$ ../scryer-prolog/target/release/scryer-prolog simsttab.pl example2.pl OK!

Example 3 failed:
$ ../scryer-prolog/target/release/scryer-prolog simsttab.pl example3.pl
?- requirements_variables(Rs, Vs),labeling([ff], Vs),print_classes(Rs).
false.
?-

[gabouelseoud]

@flexoron Thank you very much. Until triska replies, you provided me with a very good starting point. I'll try to better understand the code to find out what the problem is. I really appreciate your precious support.

[triska]

I can reproduce this! There is definitely a mistake somewhere, the only question is where exactly!

Let's break it down. First, we see that the first goal by itself already fails:

?- requirements_variables(Rs, Vs).
false.

So, adding more goals (i.e., constraints), will fail "all the more".

One thing I noticed is that we "generalize away" the last goal in requirements_variables/2, then it succeeds. To generalize away a goal, we use (*)/1 from library(debug). This is similar to "commenting out" a goal, with the advantage that it also works for the last goal in a clause, which is followed by .. So, if we change the definition of requirements_variables/2 to:

requirements_variables(Rs, Vars) :-
        requirements(Rs),
        pairs_slots(Rs, Vars),
        slots_per_week(SPW),
        Max #= SPW - 1,
        Vars ins 0..Max,
        maplist(constrain_subject, Rs),
        classes(Classes),
        teachers(Teachers),
        rooms(Rooms),
        maplist(constrain_teacher(Rs), Teachers),
        maplist(constrain_class(Rs), Classes),
        * maplist(constrain_room(Rs), Rooms).

then it succeeds.

This all worked as expected in earlier versions of Scryer Prolog. Are you interested in trying git bisect on the Scryer repository to pinpoint the commit that caused this problem? Please try it with a version of Scryer Prolog from March 2020, since the latest commit to simsttab.pl is from that time and I am certain it worked back then. For instance, you could start at 54dce9b60eff2b2c2f940554f4f9952986e6f8c6 and then bisect your way to more recent versions!

[flexoron]

git bisect drives compiling into crates hell

For example:
Compiling lexical-core-0.4.6
error[E0308]: mismatched types
lexical-core-0.4.6/src/atof/algorithm/bigcomp.rs:243:55
let nlz = den.leading_zeros().wrapping_sub(wlz) & (u32::BITS - 1); -> expected usize, found u32

After fixing those and other errors in bigcomp.rs, bigint.rs and math.rs
Next error:

Compiling markup5ever v0.8.1
error[E0603]: module export is private
markup5ever-0.8.1/build.rs:92:14
#[derive(Deserialize, Debug)] --> private module error (due to serde-1.0.127)
Cargo.toml dependency entry fixes this : serde = "1.0.123"

Finally it compiles but now I get this syntax error message, too
$ ../scryer-prolog/target/release/scryer-prolog simsttab.pl reqs.pl
caught: error(syntax_error(inconsistent_entry),use_module/1)

hmm...I'm not an rust expert but it looks like you need the rust environment of March 2020 as well when you reset to 54dce9b60eff2b2c2f940554f4f9952986e6f8c6

[triska]

Thank you a lot for looking into this!

At some point, Scryer did not yet support the discontiguous directive, this is a rather recent feature!

To prevent this syntax error, you can comment out the discontiguous directives in simsttab.pl, and then you must also ensure that all clauses in reqs.pl are actually contiguous. You can do this in Emacs by marking the hole buffer (C-x h) and then sorting all facts alphabetically with M-x sort-lines RET.

[gabouelseoud]

My point of view is that it is better to update simsttab to work correctly with the latest scryer version. You are quite a respected expert @triska. I think you can do it and this will assure us all to be able to make use of it, especially those who intend to use it as a block within a large code.

…

On Wed, 4 Aug 2021 at 23:56, Markus Triska ***@***.***> wrote: Thank you a lot for looking into this! At some point, Scryer did not yet support the discontiguous directive, this is a rather recent feature! To prevent this syntax error, you can comment out the discontiguous directives in simsttab.pl, and then you must also ensure that all clauses in reqs.pl are actually *contiguous*. You can do this in Emacs by marking the hole buffer (C-x h) and then sorting all facts alphabetically with M-x sort-lines RET. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIUQ7APVHN4NBBJROA3DUQTT3GZQDANCNFSM5BQXTGTQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

-- *===============================================* *بسم الله الرحمن الرحيم الحمد لله رب العالمين و صلاة و سلاما على الحبيب محمد رسول الله * *و اﻷنبياء أجمعين* *May Allah's Peace and Prayers be upon prophet Mohammad and his fellow prophets, Amen.* *=================================================* *اللهم ارحم والدي الحبيب و اجعله من الذين هم أحياء عند ربهم يرزقون..و بارك لي في والدتي و لا تحرمني منها أبدا* *https://www.facebook.com/abouelseoud.saleh <https://www.facebook.com/abouelseoud.saleh>* *==============================================* *Gehan A**bou**elseoud Saleh* *Phd in** El**ectrical Engineering* *Faculty of Engineering* *Alexandria University, Egypt* *=============================================================* * In the name of Allah, the Beneficent, the Merciful.**Praise be to Allah, the Cherisher and Sustainer of the Worlds. **Most Gracious, Most Merciful. **Master of the, Day of Judgment. **Thee do we worship, and Thine aid we seek. **Show us the straight way. **The way of those on whom Thou hast bestowed Thy Grace, Those whose (portion) is not wrath and who go not astray.* *بِسْمِ اللَّهِ الرَّحْمَنِ الرَّحِيمِ **الْحَمْدُ لِلَّهِ رَبِّ الْعَالَمِينَ **الرَّحْمَنِ الرَّحِيمِ **مَالِكِ يَوْمِ الدِّينِ **إِيَّاكَ نَعْبُدُ وَإِيَّاكَ نَسْتَعِينُ **اهْدِنَا الصِّرَاطَ المُسْتَقِيمَ **صِرَاطَ الَّذِينَ أَنْعَمْتَ عَلَيْهِمْ غَيْرِ المَغْضُوبِ عَلَيْهِمْ وَلاَ الضَّالِّينَ * *============================================================================* *M**a**y we all be a message of mercy and love from Allah to all His creatures.* *إنه من أكثر الأمور إيلاما في هذه الدنيا ألا نجد ممن نحبهم من استشعار آلامنا و مشاركتنا همومنا ما نرجو و الأشد من ذلك ألما أن نجد أننا مثلهم*

[flexoron]

this works, too:
reqs.pl
%room_alloc(r1, '1a', sjk, 0).
%room_alloc(r1, '1a', sjk, 1).
room_alloc(r1, '4c', mat, 0).
room_alloc(r1, '4d', mat, 0).

the issue is with:
room_alloc(r1, '1a', sjk, 0).
room_alloc(r1, '1a', sjk, 1).

[flexoron]

this works, too:
reqs.pl
room_alloc(r1, '2a', sjk, 0).
room_alloc(r1, '2a', sjk, 1).
room_alloc(r1, '4c', mat, 0).
room_alloc(r1, '4a', mat, 0).

the system does not like (to be:-) '1a'

[gabouelseoud]

The way I see it now, we are looking for a bug in the new scryer-prolog release. This is why I believe so: We have examples in the form of a set of constraints. older version of scryer-prolog successfully found variable values capable of satisfying all constraints. So it is a fact that solutions to these problems exist. However, the new scryer version fails to find these solutions. I believe that the kind of tests made by @flexoron can help some one who understands first order logic to find what constraints the new prolog fails to satisfy and guess the bug based on that. But for myself, I am new to logic. I hope @triska can help us guess what the bug is, so that it can be corrected.

…

On Thu, 5 Aug 2021 at 01:08, flexoron ***@***.***> wrote: this works, too: reqs.pl room_alloc(r1, '2a', sjk, 0). room_alloc(r1, '2a', sjk, 1). room_alloc(r1, '4c', mat, 0). room_alloc(r1, '4a', mat, 0). the system does not like (to be:-) '1a' — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIUQ7AK5RSSFXLQPCGZZUWTT3HB7TANCNFSM5BQXTGTQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

-- *===============================================* *بسم الله الرحمن الرحيم الحمد لله رب العالمين و صلاة و سلاما على الحبيب محمد رسول الله * *و اﻷنبياء أجمعين* *May Allah's Peace and Prayers be upon prophet Mohammad and his fellow prophets, Amen.* *=================================================* *اللهم ارحم والدي الحبيب و اجعله من الذين هم أحياء عند ربهم يرزقون..و بارك لي في والدتي و لا تحرمني منها أبدا* *https://www.facebook.com/abouelseoud.saleh <https://www.facebook.com/abouelseoud.saleh>* *==============================================* *Gehan A**bou**elseoud Saleh* *Phd in** El**ectrical Engineering* *Faculty of Engineering* *Alexandria University, Egypt* *=============================================================* * In the name of Allah, the Beneficent, the Merciful.**Praise be to Allah, the Cherisher and Sustainer of the Worlds. **Most Gracious, Most Merciful. **Master of the, Day of Judgment. **Thee do we worship, and Thine aid we seek. **Show us the straight way. **The way of those on whom Thou hast bestowed Thy Grace, Those whose (portion) is not wrath and who go not astray.* *بِسْمِ اللَّهِ الرَّحْمَنِ الرَّحِيمِ **الْحَمْدُ لِلَّهِ رَبِّ الْعَالَمِينَ **الرَّحْمَنِ الرَّحِيمِ **مَالِكِ يَوْمِ الدِّينِ **إِيَّاكَ نَعْبُدُ وَإِيَّاكَ نَسْتَعِينُ **اهْدِنَا الصِّرَاطَ المُسْتَقِيمَ **صِرَاطَ الَّذِينَ أَنْعَمْتَ عَلَيْهِمْ غَيْرِ المَغْضُوبِ عَلَيْهِمْ وَلاَ الضَّالِّينَ * *============================================================================* *M**a**y we all be a message of mercy and love from Allah to all His creatures.* *إنه من أكثر الأمور إيلاما في هذه الدنيا ألا نجد ممن نحبهم من استشعار آلامنا و مشاركتنا همومنا ما نرجو و الأشد من ذلك ألما أن نجد أننا مثلهم*

[gabouelseoud]

I just tried modifying the reqs file as triska described, so that lines referring to the same predicate are together and I commented out the discontinguous (files are attached) and I run using an older version of scryer that I installed using the command: cargo install scryer-prolog (instead of cargo run --release). It now works. However, I am worried about not understanding why the new version fails. There is a bug in the new engine that makes it fail to find a solution where it exists, and unless we know what it is, we may be getting wrong results. (I changed from .pl to .txt to be able to attach the files here for you to review them)
simsttab_old.txt
reqs_new.txt

[flexoron]

@triska: sorting is not an option because when sorted both scryer versions do the job.
I wonder how Example3 (https://www.metalevel.at/prolog/timetabling) works(it is not sorted)?

[gabouelseoud]

example 3 doesnot work with the new version and old version both.
Sorting works for me only with the old version after commenting out discontinguous lines as in attached files in my previous message,
NO, sorting doesn't work with the newer version, I get: Segmentation fault (core dumped)
and before that huge memory consumption that ends in fault. (I am assuming you mean by sorting the modification I did for the reqs.pl file in my version reqs_new.pl in my previous message)

[flexoron]

example3.pl is cut/paste from www.metalevel.at/prolog/timetabling

Well, here it works if sorted with this command.

$ sort example3.pl > example3-sorted.pl

$ # discontinguous disabled
$ ../scryer-prolog-old/target/release/scryer-prolog simsttab-old.pl example3-sorted.pl

$ # discontinguous enabled
$ ../scryer-prolog-new/target/release/scryer-prolog simsttab-new.pl example3-sorted.pl

[gabouelseoud]

@flexoron No, I tried your version example3-sorted.pl and got exactly my same earlier results:
old version works smoothly
new version of scryer consumes a lot of memory and ends in segmentation fault.
It appears that the newer version seeks a more (un-necessarily) memory-consuming route to solution. May be your computer has better memory resources than mine ;that's why you don't get segmentation fault. But at any event, this seems to be a bug since the old scryer finds the solution quite smoothly. I have 4 G ram