[Users] McLachlan code generation working in current stable ET release?

[Roland Haas]

Hello Bernard, all,

the failure is due to a change in ValueQ:

https://reference.wolfram.com/language/ref/ValueQ.html

which was updated in 12.2 (see "Details") and old behaviour would be
restored by using ValueQ[_, Method->"Legacy"] but I have no clue of
what that would mean.

The failure goes away if one patches Kranc (though the issue may exist
in other places as well, ValueQ seems pretty basic functionality to me):

diff --git a/Tools/CodeGen/TensorTools.m b/Tools/CodeGen/TensorTools.m
index 9e53a0a..1f17b35 100644
--- a/Tools/CodeGen/TensorTools.m
+++ b/Tools/CodeGen/TensorTools.m
@@ -301,7 +301,7 @@ DefineTensor[T_] :=
       Module[
         {c},
         c = tensorCharacter[{is}];
-        If[!ValueQ[TensorCharacter[T]],
+        If[!ValueQ[TensorCharacter[T], Method -> "Legacy"],
            TensorCharacter[T] = c,
            (* else *)
            If[!charactersMatch[c,TensorCharacter[T]],

Best to report this on the Kranc issue tracker
https://github.com/ianhinder/Kranc/issues I'd say.

Note that "Method -> Legacy" probably *breaks* on older MMA versions
(where the option may not yet exist) so is not a fix we can apply to
Kranc in that case.

Yours,
Roland

> Hello Bernard,
> 
> yes, fails for me with MMA 12.2.0 as well:
> 
> $ make McLachlan_BSSN.out
> rm -rf ML_BSSN ML_BSSN_Helper
> env ML_CODE=BSSN ./runmath.sh McLachlan_BSSN.m McLachlan_BSSN
> Using Kranc installation at ../../../repos/Kranc/Bin/..
> Mathematica 12.2.0 Kernel for Linux x86 (64-bit)
> Copyright 1988-2020 Wolfram Research, Inc.
> Profiling disabled
> 
> MapThread::mptd: Object TensorTools`Private`TensorCharacter[dir] at position {2, 2} in MapThread[NumberQ[#1] || NumberQ[#2] || #1 === #2 & , {{u}, TensorTools`Private`TensorCharacter[dir]}] has only 0 of required 1 dimensions.
> 
> MapThread::mptd: Object TensorTools`Private`TensorCharacter[epsdiss] at position {2, 2} in MapThread[NumberQ[#1] || NumberQ[#2] || #1 === #2 & , {{u}, TensorTools`Private`TensorCharacter[epsdiss]}] has only 0 of required 1 dimensions.
> 
> Riffle::list: List expected at position 1 in Riffle[TensorTools`Private`TensorCharacter[gt], ,].
> 
> StringJoin::string: String expected at position 1 in StringJoin[Riffle[TensorTools`Private`TensorCharacter[gt], ,]].
> 
> StringJoin::string: String expected at position 2 in Tensor indices in gt[la,lb] do not match those used previously: gt[<>Riffle[TensorTools`Private`TensorCharacter[gt], ,]<>].
> Error:
> StringJoin["Tensor indices in gt[la,lb] do not match those used previously: gt[", Riffle[TensorTools`Private`TensorCharacter[gt], ","], "]"]
> make: *** [Makefile:17: McLachlan_BSSN.out] Error 1
> 
> Yours,
> Roland
> 
> > Hello Bernard,
> >   
> > > Hi. I'm trying to regenerate the McLachlan ML_BSSN thorn by running the
> > > Mathematica-based scripts on my Macbook. The process is failing, and I'd
> > > like to know if it's an issue with my Mathematica installation or the
> > > McLachlan code?  
> > Being the person who last regenerated McLachlan for the last ET
> > release: I just gave this a try with 12.0.0 on Linux and there at least
> > 
> > env ML_CODE=BSSN ./runmath.sh McLachlan_BSSN.m McLachlan_BSSN
> > 
> > works.
> > 
> > I will update to 12.2.1 and see what happens. If that fails then at
> > least we know that we can check Wolframs docs do see what they
> > functionality they removed to make this fail.
> > 
> > Yours,
> > Roland
> >   
> 
> 


-- 
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://pgp.mit.edu .
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
Url : http://lists.einsteintoolkit.org/pipermail/users/attachments/20210224/8b577155/attachment.bin