Problem with Maxima 5.17.1 and ECL 8.12.0

Forgot to copy this to the maxima mailing list, so here
it is (two test failures of Maxima from CVS with Ecl 8.12.0,
and problem finding the documentation).

Meanwhile I found the problem with the apparently missing documentation
(and other broken links):

What used to be installed:
5.17.1/share/maxima/5.17.1
has now been renamed:
5.17.1/share/maxima/5.17post

This seems quite unfortunate to me: Is this only temporary?
Or what are the precise rules for the naming of this directory?
I cannot imagine any good reason to switch from the well-defined
old naming scheme to this currently undefined new naming scheme (what
is "not post"?), which just creates unnecessary trouble (now a little
script has to be written, to make the case distinction between
"post" and "non-post").

I hope this issue can be clarified.

Oliver



On Tue, Jan 06, 2009 at 04:58:56PM +0000, Oliver Kullmann wrote:
> Hello,
> 
> I got it now installing, but still problems: Some test
> failures (with Ecl), and where's the html-documentation?
> 
> The test failures are:
> 
> Running tests in rtest14:
> ********************** Problem 139 ***************
> Input:
> closeto(AS_10_4_17(- 10), 1.e-15)
> 
> 
> Result:
>                2  20 sqrt(10)             2  20 sqrt(10)
> 10 (bessel_j(- -, -----------) - bessel_j(-, -----------))
>                3       3                  3       3
> ---------------------------------------------------------- + airy_dai(- 10)
>                             3
> 
> This differed from the expected result:
> true
> 
> 293/294 tests passed (not counting 2 expected errors).
> 
> The following 1 problem failed: (139)
> 
> 
> Running tests in rtest_gamma:
> ********************** Problem 592 ***************
> Input:
> relerror(fresnel_s(%i + 5.0),
> 3.84389093777759513198736581956757227475265101347572827364567521b4 %i
>  - 2.044525505063210590493330126425293361797143144299005524393297869b5, 2.e-14)
> 
> 
> Result:
> 2.542490092925679718842752779534059543174617462409733517737362792b-14
> 
> This differed from the expected result:
> true
> 
> ********************** Problem 614 ***************
> Input:
> relerror(fresnel_c(%i + 5.0),
> 2.044530505063119004499308897858846121088557663560705550579996093b5 %i
>  + 3.84394093777740143202918713550472184235160647415045418329908291b4, 2.e-14)
> 
> 
> Result:
> 2.545880046763412762594157711951056394921468040249292492237741923b-14
> 
> This differed from the expected result:
> true
> 
> 614/616 tests passed.
> 
> The following 2 problems failed: (592 614)
> 
> -------------
> 
> Regarding the documentation: It used to be in
> 
> 5.17.1/share/maxima/5.17.1/doc/
> 
> but this is now (using the CVS-maxima version and boostrap)
> apparently no longer build?
> 
> Oliver
> 
> 
> 
> On Tue, Dec 30, 2008 at 01:58:38PM -0700, Robert Dodier wrote:
> > Looking at the CVS logs for src/numth.lisp and src/rat3c.lisp,
> > I see that the $JACOBI stuff was applied after 5.17.1 was tagged.
> > However those changes have been applied to both the 5.17 branch
> > (on Dec 21) and the main trunk (on Dec 25).
> > 
> > On 12/30/08, Oliver Kullmann <O.Kullmann at swansea.ac.uk> wrote:
> > 
> > > Some more attempts:
> > >
> > >  Since I don't know how to update the 5.17.1-distribution
> > >  (it doesn't match what you find in CVS) I did the following:
> > 
> > I can't tell what would be the outcome of the steps which you described.
> > My advice is to start clean (do not try to update or modify an
> > existing installation) and try one of the following:
> > 
> > 5.17 branch:
> > 
> > $ cvs -d<whatever> co -rRELEASE-5_17-BRANCH maxima
> > 
> > CVS head:
> > 
> > $ cvs -d<whatever> co maxima
> > 
> > and then
> > 
> > $ sh bootstrap
> > $ LDFLAGS=-Wl,-rpath=/usr/local/lib ./configure --enable-ecl
> > $ make
> > 
> > It might be possible to omit LDFLAGS, I don't know.
> > 
> > Hope this helps,
> > 
> > Robert
>