From Christian.Maeder at dfki.de Mon Mar 16 12:00:58 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Mon, 16 Mar 2009 12:00:58 +0100 Subject: [Hets-devel] [Hets-users] Problems with existential equations In-Reply-To: <429c032d0903132202j52059467l3b20c6272b2019a5@mail.gmail.com> References: <429c032d0903132202j52059467l3b20c6272b2019a5@mail.gmail.com> Message-ID: <49BE316A.70008@dfki.de> Hi, Yes, Isabelle is supposed to parse "=e=" via the imported file "$HETS_LIB/Isabelle/MainHCPairs" or "$HETS_LIB/Isabelle/MainHC" which is part of the libraries in Hets-lib. https://svn-agbkb.informatik.uni-bremen.de/Hets-lib/trunk/ Set HETS_LIB to point to your Hets-lib directory. Didn't you get an error? *** Could not find theory file "MainHCPairs.thy" in ".", ... If not, how did you invoke Isabelle? Cheers Christian Michael Chan wrote: > Hi, > > I'm trying to use Isabelle to run some proofs on Basic/Numbers, in > particular I'm trying to prove the generated Numbers_Int_E1.thy. It > seems that some of the axioms might be incorrect, causing type > unification to fail. For example: > > (line 302) > ga_projection_injectivity [rule_format] : > "ALL x. ALL y. gn_proj_Int_Nat(x) =e= gn_proj_Int_Nat(y) --> x = y" > > Is Isabelle supposed to be able to parse =e=? The error I'm getting is: > > *** Type unification failed: Clash of types "*" and "bool" > *** Type error in application: Incompatible operand type > *** > *** Operator: op = (gn_proj_Int_Nat(x) = e) :: bool => bool > *** Operand: gn_proj_Int_Nat(y) :: bool * X_Nat > *** > *** The error(s) above occurred in axiom "ga_projection_injectivity" > *** At command "axioms". > > It seems =e= isn't properly parsed. From Christian.Maeder at dfki.de Mon Mar 16 13:15:12 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Mon, 16 Mar 2009 13:15:12 +0100 Subject: [Hets-devel] [Hets-users] Using Hets with Isabelle In-Reply-To: <12d162d00903142106ve72a5cfle30930f705e8cfd5@mail.gmail.com> References: <12d162d00903141101i7a710402md8ee3f7930ce28e9@mail.gmail.com> <12d162d00903142106ve72a5cfle30930f705e8cfd5@mail.gmail.com> Message-ID: <49BE42D0.8080008@dfki.de> Stephy Wong wrote: > Also, I have one more question: How do proof obligations get discharged > with Isabelle? After completing the proof in emacs and exiting, the > status shown in the Hets interface is changed to "No Prover Running". > Should the targeted red node be expected to turn green? After exiting emacs you should return to the prover window, that indicates by a "+" which theorems have been proven by Isabelle (provided you've chosen Isabelle as prover before). Only if all theorems have this "+" the red node will turn to green after closing the prover window. (Everything else is a bug and should be reported.) > 1. Is it good style to add lemmas into the thy file to prove a > theorem? Should the supposedly lemmas be translated by Hets as > axioms instead? So basically, should the user just step through the > proof in Isabelle without modifying the generated theory? Additional lemmas are supposed to be added to the theory file, but surely the original theory may be modified to include a useful implied sentence, too. (This is a sentences that is followed by %implied and needs to be proven, too.) But prover-dependent auxiliary lemmas are not supposed to be added at will to a specification that should remain succinct/readable. In particular the basic library is not supposed to be changed (without good reason). Definitions, axioms and theorems must not be changed in the theory and also the line 'ML "Header.record.."' should follow your proof. This line ensures that hets is notified about a successful proof! Axioms may be deleted or theorems may be added, though. > 2. Should all open proof obligations in the CASL specs that come > with Hets be already provable? For example, those in > Basic/Numbers.casl and Basic/MachineNumbers. It seems that > ga_comm___XPlus___2 in Numbers and dichotomy_TotalOrder in > MachineNumbers aren't already provable. Is something missing? Many proofs for the CASL specs are missing, although we belief that the theorems hold. For you this means, that the proof chain may be incomplete. Commutativity of __+__ and dichotomy of __<=__ over Numbers should be provable, but it may be difficult to actually do so. (Or do you have a counter example?) For your own proofs you should simply assume that these properties hold, but we would gladly accept proofs for our basic library, too. Cheers Christian From dayzman at gmail.com Mon Mar 16 13:20:31 2009 From: dayzman at gmail.com (Michael Chan) Date: Mon, 16 Mar 2009 12:20:31 +0000 Subject: [Hets-devel] [Hets-users] Problems with existential equations In-Reply-To: <49BE316A.70008@dfki.de> References: <429c032d0903132202j52059467l3b20c6272b2019a5@mail.gmail.com> <49BE316A.70008@dfki.de> Message-ID: <429c032d0903160520t70bc9433j7df7be154d1da3e4@mail.gmail.com> Hi Christian, I have $HETS_LIB properly set and $HETS_LIB/Isabelle/MainHCPairs exists, so I don't get the error you mentioned. I invoke Isabelle by right clicking on a red node, then Prove, then selecting Isabelle, and then Prove. Strangely, if I load the theory in a shell, the =e= gets parsed. So the problem occurs only when it's invoked via Hets. Thanks. Michael On Mon, Mar 16, 2009 at 11:00 AM, Christian Maeder wrote: > Hi, > > Yes, Isabelle is supposed to parse "=e=" via the imported file > "$HETS_LIB/Isabelle/MainHCPairs" or "$HETS_LIB/Isabelle/MainHC" > which is part of the libraries in Hets-lib. > > https://svn-agbkb.informatik.uni-bremen.de/Hets-lib/trunk/ > > Set HETS_LIB to point to your Hets-lib directory. > > Didn't you get an error? > *** Could not find theory file "MainHCPairs.thy" in ".", ... > > If not, how did you invoke Isabelle? > > Cheers Christian > > > Michael Chan wrote: > > Hi, > > > > I'm trying to use Isabelle to run some proofs on Basic/Numbers, in > > particular I'm trying to prove the generated Numbers_Int_E1.thy. It > > seems that some of the axioms might be incorrect, causing type > > unification to fail. For example: > > > > (line 302) > > ga_projection_injectivity [rule_format] : > > "ALL x. ALL y. gn_proj_Int_Nat(x) =e= gn_proj_Int_Nat(y) --> x = y" > > > > Is Isabelle supposed to be able to parse =e=? The error I'm getting is: > > > > *** Type unification failed: Clash of types "*" and "bool" > > *** Type error in application: Incompatible operand type > > *** > > *** Operator: op = (gn_proj_Int_Nat(x) = e) :: bool => bool > > *** Operand: gn_proj_Int_Nat(y) :: bool * X_Nat > > *** > > *** The error(s) above occurred in axiom "ga_projection_injectivity" > > *** At command "axioms". > > > > It seems =e= isn't properly parsed. > -------------- next part -------------- An HTML attachment was scrubbed... URL: From Christian.Maeder at dfki.de Mon Mar 16 13:29:49 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Mon, 16 Mar 2009 13:29:49 +0100 Subject: [Hets-devel] [Hets-users] Problems with existential equations In-Reply-To: <429c032d0903160520t70bc9433j7df7be154d1da3e4@mail.gmail.com> References: <429c032d0903132202j52059467l3b20c6272b2019a5@mail.gmail.com> <49BE316A.70008@dfki.de> <429c032d0903160520t70bc9433j7df7be154d1da3e4@mail.gmail.com> Message-ID: <49BE463D.1010305@dfki.de> Michael Chan wrote: > Strangely, if I load the theory in a shell, the =e= gets parsed. So the > problem occurs only when it's invoked via Hets. Hets simply calls "Isabelle" or the binary pointed to by the environment variable HETS_ISABELLE plus the theory file as argument. Which OS are you using? Do you have blanks (or other funny characters) in your paths? Cheers Christian From dayzman at gmail.com Mon Mar 16 13:47:33 2009 From: dayzman at gmail.com (dayzman at gmail.com) Date: Mon, 16 Mar 2009 12:47:33 +0000 Subject: [Hets-devel] [Hets-users] Problems with existential equations In-Reply-To: <49BE463D.1010305@dfki.de> Message-ID: <0016e6d7eaacaffaef04653bda7f@google.com> On Mar 16, 2009 12:29pm, Christian Maeder wrote: > Michael Chan wrote: > > Strangely, if I load the theory in a shell, the =e= gets parsed. So the > > problem occurs only when it's invoked via Hets. > Hets simply calls "Isabelle" or the binary pointed to by the environment > variable HETS_ISABELLE plus the theory file as argument. Which OS are > you using? Do you have blanks (or other funny characters) in your paths? I've just checked out the Hets-lib from the trunk, and now it all seems to work! The culprit seems to be Isabelle/MainHCPairs.thy. FYI, I'm using Linux 64bits here. Thanks a lot! Michael > Cheers Christian -------------- next part -------------- An HTML attachment was scrubbed... URL: From Christian.Maeder at dfki.de Wed Mar 18 18:58:34 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Wed, 18 Mar 2009 18:58:34 +0100 Subject: [Hets-devel] [Hets-users] Compiling Hets In-Reply-To: <12d162d00903181050y6c08724ev6da885cb5a862ff4@mail.gmail.com> References: <12d162d00903170921p4254f493yd3c4782a3fdb9861@mail.gmail.com> <49BFD289.4030000@dfki.de> <12d162d00903181050y6c08724ev6da885cb5a862ff4@mail.gmail.com> Message-ID: <49C1364A.3000103@dfki.de> Stephy Wong wrote: > I've managed to install the packages, but I can't seem to compile uni. I > get the following from 'make packages': Could you try "make cabal" after ./configure and make sure that "ghc-pkg list" does not list any uni packages. (remove them using: ghc-pkg unregister uni- --user) > ghc-6.8.2: /home/swong/uni/util/uni-util.o: unknown symbol > `networkzm2zi2zi0zi1_NetworkziBSD_a34_closure' > Loading package uni-util ... linking ... ghc-6.8.2: unable to load > package `uni-util' > make[1]: *** [libfasthere] Error 1 > make[1]: Leaving directory `/home/swong/uni/posixutil' > make: *** [packages] Error 2 I'm not sure, what goes wrong here. Cheers Christian From Christian.Maeder at dfki.de Thu Mar 19 09:45:12 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Thu, 19 Mar 2009 09:45:12 +0100 Subject: [Hets-devel] Compiling Hets In-Reply-To: <001636eefc77357a21046568abab@google.com> References: <001636eefc77357a21046568abab@google.com> Message-ID: <49C20618.7050600@dfki.de> s.wong.731 at googlemail.com wrote: > /home/swong/uni/mk/cabal-ghc --prefix=/home/swong/uni --user; runhaskell > /home/swong/uni/Setup.hs build; runhaskell /home/swong/uni/Setup.hs > copy; runhaskell /home/swong/uni/Setup.hs register --gen-script; > ./register.sh ; fi > /bin/sh: ./register.sh: not found > make[1]: *** [cabalhere] Error 127 > make[1]: Leaving directory `/home/swong/uni/util' > make: *** [cabal] Error 2 You seem to have old uni sources. uni-ghc-6.8.x-HaXml-1.13.2-2008-01-25.tgz is too old. The "--gen-script" stuff has been outdated. Please check uni out from svn: svn co https://svn-agbkb.informatik.uni-bremen.de/uni/trunk uni Or get the current sources (I've just put up) from http://www.informatik.uni-bremen.de/agbkb/forschung/formal_methods/CoFI/hets/src-distribution/versions/uni-2009-03-19.tgz Christian From Christian.Maeder at dfki.de Thu Mar 26 12:32:32 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Thu, 26 Mar 2009 12:32:32 +0100 Subject: [Hets-devel] release 0.93 preparation Message-ID: <49CB67D0.7040102@dfki.de> Dear Hets-Developers, The next release is due tomorrow: http://trac.informatik.uni-bremen.de:8080/hets/roadmap Many tickets are still open and will remain so, yet it is important to create a stable release integrating the improvements made since the last release Oct-2008. http://trac.informatik.uni-bremen.de:8080/hets/wiki/ReleaseNotes I call a hets binary "stable", if it passes all the tests that I run nightly and the interactive tests behave as described http://trac.informatik.uni-bremen.de:8080/hets/wiki/InteractiveTests The cron-jobs for the night are checked in as Hets-sources: Hets/utils/nightly/cronjob.sh Hets/utils/nightly/{intel-mac,linux,mac,pc-solaris,solaris}/cronjob.sh Not all tests say if they "passed", so the resulting log needs manual inspection. More automatic tests would be desirable particular for logics that I'm not familiar with. A big burden are interactive tests or "playing" with the GUI to detect problems. To this regard the GUI is not stable at all and may crash in the middle of proving, but I see no chance to improve the situation other than getting reports from users. And even for bugs like http://trac.informatik.uni-bremen.de:8080/hets/ticket/646 we'll have a hard time. Yet, I'll be happy to receive suggestions (or better yet a hand) on improving the next release -- or futures releases. Cheers Christian From Till.Mossakowski at dfki.de Thu Mar 26 12:49:11 2009 From: Till.Mossakowski at dfki.de (Till Mossakowski) Date: Thu, 26 Mar 2009 12:49:11 +0100 Subject: [Hets-devel] release 0.93 preparation In-Reply-To: <49CB67D0.7040102@dfki.de> References: <49CB67D0.7040102@dfki.de> Message-ID: <49CB6BB7.2000801@dfki.de> Christian Maeder schrieb: > A big burden are interactive tests or "playing" with the GUI to detect problems. > To this regard the GUI is not stable at all and may crash in the middle of > proving, but I see no chance to improve the situation other than getting reports > from users. Maybe Razvan could add a command to the command-line interface that would keep running uDrawGraph in the background (but without blocking the further processing of the script). Most GUI commands are also available as script commands. Then we could use scripts to test the GUI. In order to be really able to use this within batch processing, uDrawGraph should provide some test mode that does not actually use a graphical display, but that produces the same messages as in normal mode. All the best, Till -- Till Mossakowski Cartesium, room 2.051 Phone +49-421-218-64226 DFKI GmbH Bremen Fax +49-421-218-9864226 Safe & Secure Cognitive Systems Till.Mossakowski at dfki.de Enrique-Schmidt-Str. 5, D-28359 Bremen http://www.dfki.de/sks/till Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH principal office, *not* the address for mail etc.!!!: Trippstadter Str. 122, D-67663 Kaiserslautern management board: Prof. Wolfgang Wahlster (chair), Dr. Walter Olthoff supervisory board: Prof. Hans A. Aukes (chair) Amtsgericht Kaiserslautern, HRB 2313 From Christian.Maeder at dfki.de Thu Mar 26 13:39:27 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Thu, 26 Mar 2009 13:39:27 +0100 Subject: [Hets-devel] release 0.93 preparation In-Reply-To: <49CB6BB7.2000801@dfki.de> References: <49CB67D0.7040102@dfki.de> <49CB6BB7.2000801@dfki.de> Message-ID: <49CB777F.5000503@dfki.de> Till Mossakowski wrote: > Christian Maeder schrieb: > >> A big burden are interactive tests or "playing" with the GUI to detect >> problems. >> To this regard the GUI is not stable at all and may crash in the >> middle of >> proving, but I see no chance to improve the situation other than >> getting reports >> from users. > > Maybe Razvan could add a command to the command-line interface that > would keep running uDrawGraph in the background (but without blocking > the further processing of the script). Most GUI commands are also > available as script commands. Then we could use scripts to test > the GUI. In order to be really able to use this within batch processing, > uDrawGraph should provide some test mode that does not actually > use a graphical display, but that produces the same messages as > in normal mode. Good idea, but I would already be happy if scripts alone would go through. http://trac.informatik.uni-bremen.de:8080/hets/ticket/680 Christian From Christian.Maeder at dfki.de Tue Mar 31 13:42:28 2009 From: Christian.Maeder at dfki.de (Christian Maeder) Date: Tue, 31 Mar 2009 13:42:28 +0200 Subject: [Hets-devel] hets-0.93 released Message-ID: <49D201A4.3090708@dfki.de> Dear Hets friends, we proudly present our new release 0.93 of the Heterogeneous Tool Set (Hets) http://www.dfki.de/sks/hets The features can be looked up at: http://trac.informatik.uni-bremen.de:8080/hets/wiki/ReleaseNotes Installers or binaries can be downloaded from: http://www.dfki.de/sks/hets/installation_e.htm (A live-CD will follow soon) Enjoy it, Christian -- _______________________________________________________________ Dr. Christian Maeder Hausadresse: DFKI Bremen Universit?t Bremen Sichere Kognitive Systeme Cartesium 02.053 Robert-Hooke-Str. 5 Enrique-Schmidt-Str. 5 28359 Bremen 28359 Bremen http://www.dfki.de/sks www.informatik.uni-bremen.de/~maeder Telefon: +49 (421) 218-64227 Telefax: +49 (421) 218-9864227 E-Mail: Christian.Maeder at dfki.de _______________________________________________________________ Deutsches Forschungszentrum f?r K?nstliche Intelligenz GmbH Firmensitz: Trippstadter Stra?e 122, D-67663 Kaiserslautern Gesch?ftsf?hrung: Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender) Dr. Walter Olthoff Vorsitzender des Aufsichtsrats: Prof. Dr. h.c. Hans A. Aukes Amtsgericht Kaiserslautern, HRB 2313