segfault on gdal.OpenEx() call

I guess you're using the gdal packages from the UbuntuGIS PPA, since Ubuntu itself doesn't have 2.1.0.

Can you provide a link to the IRC discussion or otherwise provide more motivation why the DODS driver should be disabled now. It has been enabled for a very long time, so I'm surprised Even now recommends disabling it.

See:
http://irclogs.geoapt.com/osgeo/%23osgeo.2016-11-09.log

Op 9 nov. 2016 07:30 schreef "Bas Couwenberg" <email address hidden>:

> I guess you're using the gdal packages from the UbuntuGIS PPA, since
> Ubuntu itself doesn't have 2.1.0.
>
> Can you provide a link to the IRC discussion or otherwise provide more
> motivation why the DODS driver should be disabled now. It has been
> enabled for a very long time, so I'm surprised Even now recommends
> disabling it.
>
> --
> You received this bug notification because you are a member of
> UbuntuGIS, which is subscribed to gdal in Ubuntu.
> https://bugs.launchpad.net/bugs/1640360
>
> Title:
> segfault on gdal.OpenEx() call
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/gdal/+bug/1640360/+subscriptions
>

00:24:09 transit: Hi, can I get a mantra so I can create an OSGeo account to register a bug report?
00:26:53 EvenR: transit: can you very shortly tell aobut the bug ?
00:28:45 transit: SegFault: https://dpaste.de/ppUM both libgdal20 & python-gdal at v2.1.0 on Ubuntu.
00:28:46 sigq: Title: dpaste.de: Snippet #389262 (at dpaste.de)
00:30:02 EvenR: transit: mantra givein in private chat. Hum I suspect this is an issue with the DODS driver and libdap that tend to segfault on non expected URLs
00:30:13 EvenR: you can try to define GDAL_SKIP=DODS as environment variable
00:32:02 transit: EvenR: That eliminates the segfault.
00:32:14 transit: EvenR: Still worth reporting?
00:33:54 EvenR: transit: perhaps report to the packagers. I'd recommend them to disable the DODS driver in their builds

That's not very informative.

I've inquired about the state of the DODS driver, before I can decided to disable it or not:

 https://lists.osgeo.org/pipermail/gdal-dev/2016-November/045475.html

Even has committed a fix for the segfault, and I've included it as a patch in the Debian package, see the discussion on the GDAL mailinglist:

 https://lists.osgeo.org/pipermail/gdal-dev/2016-November/045477.html

New revisions of the gdal package for trusty & xenial will be available in the ubuntugis-unstable PPA shortly.

Fixed packages are available in Debian unstable and the various UbuntuGIS PPAs.

New package eliminates segfault, thanks :-)

One change I see with the new package is the KML driver reporting its short name as "LIBKML" instead of "KML". Is this a problem?

That wasn't changed between 2.1.0+dfsg-1~xenial0 & 2.1.0+dfsg-1~xenial1.

There's no change regarding KML. Both drivers KML and LIBKML exist. There LIBKML is normally the first one to be tried when iterating over drivers, hence the LIBKML name being reported

Ok, thanks for your time guys.