Re: [Reduce-algebra-developers] tr in csl

[Marek K. <mar...@sa...>]

Dear Arthur,

Your recent commits (3894 - 3897) solved two of the CSL problems I
reported last week:
1. tracing segfault
2. catastrophic error in GCD

Thank you very much for your effort!
Marek Kosta

On 08.02.2017 12:52, Arthur Norman wrote:
> I am busy right now re-working the trace mechanism in a manner where the
> implementation will be internally much tidier and so in a way that I hope and
> expect to make it more reliable. The current effect is fall-out from the
> transition to use of C++ catch and throw and to using class destructors to reset
> stack pointers and restore fluid variable (etc) when functions are exited. It
> makes more sense to me that I move forward on this and complete the new and
> nicer implementation rather than delay that by patching something that I believe
> only has days to remain in place! It remain possible for people to use a
> slightly earlier revision of CSL when this is a bad disruption to them or use
> PSL until I have things fixed. Which - as I say - I am right in the middle of!
> 
> Arthur
> 
> On Wed, 8 Feb 2017, Marek Kosta wrote:
> 
>> Dear all,
>>
>> tracing in CSL produces segfaults. Consider the following input:
>>
>> --- BEGIN ---
>> rlset r$
>> on ofsfvs$
>> tr vsdb_expandNode;
>> rlqe ex({k64, k65}, k43*k64**2 + 2*k43*k64*k65 + k43*k65**2 + k64**2*k65 +
>> k64*k65**2 = 0);
>> --- END ---
>>
>> running this I obtain:
>>
>> mkosta@alfred:/local/var/tmp/mkosta/reduce-algebra [524] ./bin/redcsl
>> Reduce (Free CSL version, revision 3890), 08-Feb-17 ...
>>
>> 1: rlset r$
>>
>> 2: on ofsfvs$
>>
>> 3: tr vsdb_expandNode;
>>
>> (vsdb_expandnode)
>>
>> 4: rlqe ex({k64, k65}, k43*k64**2 + 2*k43*k64*k65 + k43*k65**2 + k64**2*k65 +
>> k64*k65**2 = 0);
>> Entering vsdb_expandnode (2 args)
>> Arg1: %(vsdb (k65 k64) #1=(or (equal (((k64 . 1) . 1) ((k65 . 1) . 1)) nil) (
>> equal (((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)) ((k64 . 1) ((k65 . 1) . 1)))
>> nil)) nil (k64 k65) nil nil nil nil (vsht ((#1# #1#))) nil)
>> Arg2: (vsnd nil nil (k65 k64) (or (equal (((k64 . 1) . 1) ((k65 . 1) . 1)) nil)
>> (equal (((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)) ((k64 . 1) ((k65 . 1) . 1)))
>> nil)) nil)
>> = (nil)
>> Entering vsdb_expandnode (2 args)
>> Arg1: [vsdb (k65 k64) (or (equal (((k64 . 1) . 1) ((k65 . 1) . 1)) nil) (equal (
>> ((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)) ((k64 . 1) ((k65 . 1) . 1))) nil))
>> nil (k64 k65) nil nil nil nil (vsht ((true true) ((or (equal (((k64 . 1) . 1) ((
>> k65 . 1) . 1)) nil) (equal (((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)) ((k64 .
>> 1) ((k65 . 1) . 1))) nil)) (or (equal (((k64 . 1) . 1) ((k65 . 1) . 1)) nil) (
>> equal (((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)) ((k64 . 1) ((k65 . 1) . 1)))
>> nil))))) nil]
>> Arg2: (vsnd nil (vsts k65 (vstp (1) nil nil (vspr (((k65 . 1) . 1) ((k64 . 1) .
>> 1)) k65 ((1 . 1)) true))) (k64) true (vsnd nil nil (k65 k64) (or (equal (((k64 .
>> 1) . 1) ((k65 . 1) . 1)) nil) (equal (((k43 . 1) ((k64 . 1) . 1) ((k65 . 1) . 1)
>> ) ((k64 . 1) ((k65 . 1) . 1))) nil)) nil))
>> Segmentation fault
>>
>>
>> When commenting the "tr" command everything runs smoothly. The PSL
>> version can run with "tr" and also without "tr". As usual I tested
>> this on two independent Linux systems.
>>
>> Best,
>> Marek Kosta
>>
>> ------------------------------------------------------------------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
>> _______________________________________________
>> Reduce-algebra-developers mailing list
>> Red...@li...
>> https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers
>>