Re: [PyGiNaC-users] strange behavior with the test suite

[Jonathan B. <jbr...@ea...>]

On Mon, 2005-10-10 at 19:13 +0300, Matti Peltom=E4ki wrote:
> Hi,=20
>=20
> On Tue, 4 Oct 2005, Jonathan Brandmeyer wrote:
>=20
> > In the current fix, if the test fails one way, it is retried with the
> > second option.
>=20
> Based on Richard Kreckel's insight on what the problem may be on=20
> ginac-list, I propose to work around this such that all the exam_* test=20
> suites are run before the check_* ones, since the latter ones are those
> who introduce a random number of new symbols and affect the canonical=20
> ordering giving the irritating results we had.
>=20
> This workaround also has the advantage that it more precisely mimicks=20
> GiNaC's test suites' behavior, which, I guess, in some loose sense, is ou=
r=20
> goal.

My original goal was to use the ability to run a similar test suite as a
measure of PyGiNaC's completeness.  So, I'm not particularly concerned
about relatively minor differences like this.

>  I have committed these changes to the CVS.=20

> On the other hand, I'm quite puzzled with your changes in exam_lsolve
> between revisions 1.1 and 1.2. From my point of view the tests passed
> cleanly before tha changes but not after. Is it the other way around for
> you?

Yes it was.

>  If so, do we have some differences in our environments?

Probably.  Upon investigation, I found that str(numeric(1)) =3D=3D '1.0' on
my Sid machine and '1' on my Sarge machine.  So I changed that
particular test to operate exactly like the others in that file.  It
should work fine in both environments now.

-Jonathan