Re: probably a bug: gdal, nio, seg fault

From: Mary Haley <haley_at_nyahnyahspammersnyahnyah>
Date: Tue Aug 10 2010 - 10:24:47 MDT

Hi Mikhail,

We discussed this problem, and forgot to get back to you.

Since it works under 1.3.0b5, I'm hoping this means the problem
fixed itself, but we certainly didn't do anything specific to this.

PyNIO 1.3.0b1 didn't have any GDAL dependencies, so we
didn't need to link the GDAL libs for 1.3.0b1.

One thing that's different between 1.3.0b1 and 1.3.0b5 is
that we now do have shapefile support, and hence the GDAL
libs are needed. Also, the PyNIO source is now stand-alone, and
you don't need to link in all the NCL libraries. It's possible that
one of these NCL libraries might have had an internal routine
that GDAL picked up, instead of its own.

The next version of PyNIO will actually be 1.4.0, and we'll be
providing full source code so you don't need to install NCL
first in order to build it. We will also provide binaries for various
systems. Do you want to try any of these out this week?


On Aug 5, 2010, at 9:14 AM, Mikhail Itkin wrote:

> Hi,
> Just figured out that Gdal and Nio libraries invoked in the same script
> in the "right" order will cause a segmentation fault.
> On our servers we have Nio 1.3.0b1 installed along with Gdal 1.6.1 and
> 1.6.2.
> * If in Python script Nio is imported *after* Gdal it will cause a
> segmentation fault on opening a netcdf file.
> * If Nio is imported *prior* to importing gdal, then problem does not
> appear.
> The problem does not happen on my laptop where Nio 1.3.0 beta5 is installed.
> All software is for 64 bits.
> Cheers,
> Mikhail
> _______________________________________________
> pyngl-talk mailing list
> List instructions, subscriber options, unsubscribe:

pyngl-talk mailing list
List instructions, subscriber options, unsubscribe:
Received on Tue Aug 10 10:24:54 2010

This archive was generated by hypermail 2.1.8 : Fri Aug 13 2010 - 15:07:13 MDT