[PROJ] PROJ 6 testntv2 fails

[Greg Troxel]

Markus Metz <markus.metz.giswork at gmail.com> writes:

> I was reporting as a user compiling proj from source, not as a packager. I

Understood, but manual intervention in running tests is not scalable.

> agree that tests must succeed if only the minimum required datum grids are
> installed (why are these not included in the proj source code?) and also if
> all other datum grids available from https://proj4.org/download.html are
> installed. Additional grids e.g. mentioned in
> https://proj4.org/resource_files.html#external-resources should not be
> considered by the tests.
>
> In short, tests must use expected input and ignore unexpected input in
> order to succeed. That's what you suggest, right?

I don't follow the previous sentence.

I meant:

  If it's ok not to have the extra grids installed, and if it's
  preferred to have them (if you use those projections, then

  tests should succeed either way, somehow (by accomodating both
  answers, or by forcing proj not to use the extra grids during tests)


I suspect that mine is a minority view, so I don't want to belabor it,
beyond clearly stating it, which I hope I finally did...