psVectorClipFitPolynomial1D inverted test on stats->min,max

[eugene]

in the eam_rel8_b1 version of psMinimize.c, there is a bug in
psVectorClipFitPolynomial1D: the tests for stats->min and stats->max (to see if
they have been set) chooses the wrong option: if (isfinite(stats->min)) then
the code should use the value of stats->min not the value of stats->clipSigma.
The same error occurs for stats->max.

I've implemented the fix in eam_rel8_b2

[robert.desonia@…]

joined eam_rel8_b1 and eam_rel8_b2 into the CVS main trunk. This caused many problems in psCoord, as
George also made changes to this in response to bugs filed recently. Can you tell me if the code in your
branch fixed these issues already?

[eugene]

the changes I made fixed the linear transform inversion, and it looks like
george's code fixes the same matrix inversion bug. my code also fixed psProject
and psDeproject, which the 1.93 version does not address. On the other hand,
George added in some ASSERT and other critical tests, but I suppose those don't
cause conflicts?

[eugene]

the primary bug here (stats->min, stats->max) was still there in rel9_0

I've fixed it in eam_rel9_b0

[gusciora@…]

This code is fixed (current CVS tree).