[Shapelib] [gdal-dev] shapelib 1.6.1RC2 available (was Re: shapelib 1.6.1RC1 available)

[Greg Troxel]

Even Rouault <even.rouault at spatialys.com> writes:

>> 2) The tests all fail because they have #! lines with /bin/bash.
> Should be addressed per https://github.com/OSGeo/shapelib/pull/150

Great, thanks.

>> With the bash problem worked around, most tests pass, except
>>
>>    FAIL: tests/shpproj.sh
>>    ======================
>>
>>    ../test-driver: ./tests/shpproj.sh: not found
>>    FAIL tests/shpproj.sh (exit status: 127)
>>
>> and I don't see that in the tarball, so I'm guessing that isn't a
>> problem only I am seeing.
>>
>> shapelib doesn't depend on proj, so perhaps there needs to be
>> documentation of tests dependencies?
> I don't reproduce that error (shpproj.sh used to test a shpproj.c
> utility that was removed during the PROJ6 transition, so this is a
> remains of a more complete tests, but it has the virtue of testing a
> few of the contrib/ utilities that have no tests otherwise)

It was another /bin/bash problem.  I commented in the PR, but with
replacing the #! line, I got a clean pass.

>> 5) There are a surprisingly large number of programs in bin.  The pkgsrc
>> package didn't used to install them, but they look like that likely
>> won't conflict with other pacakges..  Do you think users of shapelib
>> expect all of these to be installed?
>
> No opinion here. I hardly use myself any of the shapelib utils. I see
> Debian ships all of them
> (https://packages.debian.org/sid/amd64/shapelib/filelist), but some of
> them come from the contrib/ area (which doesn't necessarily means they
> are worse qualities than other non-contrib utilities). I came much
> later to the party, so I don't have much historical background to
> provide.

I suggest not installing contrib stuff.  That's what I'm doing in the
package for now.

https://github.com/OSGeo/shapelib/issues/151

> Nothing of the above seems to be 1.6.1 specific, so I don't think it
> deserves a new RC

Agreed; none of this seems new in 1.6.1 vs 1.6.0.  It's just that an RC
provokes a full reexamination and report, which I hope is helpful on
balance.  With the workarounds I have in place, it seems to be fine.