[postgis-devel] [Internet] Re: Postgis 2.4.6 regession tests fails on Solaris

[Greg Troxel]

> I have tested with postgis 2.5.1 on SPARC with the same checks failing
> among new ones.

Do you mean sparc (32-bit CPU), or sparc64?

> I also tested on X86_64 with version 2.3.8, 2.4.6 and 2.5.1. All three
> checks are successful. It seems the failures comes from the
> micro-architecture. I wonder how can so few checks fails if there is
> really a major issue with SPARC.

If only a few things are off, then I guess it isn't major!   But
seriously, code is mostly ok and compilers mostly are correct, so this
situation is not entirely surprising.

The two likely suspects are:

  code that isn't right that happens to work on x86

  compiler bug

plus aways we have

  something else I didn't think of

> I also tested with geos 3.6.3 with again the same checks failing on SPARC.

It seems somewhat likely then that the root cause is in geos, assuming I
understand you correctly and that the postgis failing tests sort of
correspond to the geos failing tests.