Pete,
I guess the first step should be raising the issue here
https://github.com/Unidata/thredds/issues
and referencing @lesserwhirls (Sean Arms at Unidata) for his comment?
-Rich
On Wed, Aug 31, 2016 at 9:04 PM, <Peter.Warren@xxxxxxxx> wrote:
> Hi Rich,
>
> Thanks for the response.
>
> Assuming the Unidata guys are ok with the proposed solution, I'll setup a
> pull request for ncwms and edal-java.
>
> I notice in edal-java code there is a TODO stating that the default reading
> strategy chooser should take into account the size of the source grid. I had
> a brief look at implementing this but it's non-trivial for complicated grids.
>
> Cheers
> Pete
>
> -----Original Message-----
> From: Signell, Richard [mailto:rsignell@xxxxxxxx]
> Sent: Thursday, 1 September 2016 12:07 AM
> To: Warren, Peter (Mineral Resources, North Ryde) <Peter.Warren@xxxxxxxx>
> Cc: THREDDS community <thredds@xxxxxxxxxxxxxxxx>
> Subject: Re: [thredds] NetCDF-4 DataReadingStrategy
>
> Peter,
> The THREDDS Data Server code is now here:
> https://github.com/Unidata/thredds/
> so you could fork it and submit a pull request for this change if it hasn't
> already been handled in ncWMS2, which will be coming out with TDS 5.0, I
> think in October?
>
> Having said this, I'm sure the Unidata guys will now chime in here to
> correct me. ;-)
>
> -RIch
>
> On Wed, Aug 31, 2016 at 1:56 AM, <Peter.Warren@xxxxxxxx> wrote:
>> Hi Thredds List,
>>
>>
>>
>> My colleagues and I (at CSIRO and Geoscience Australia) have been
>> using a patched version of thredds 4.2 for over 5 years to deliver
>> satellite earth observation data. We offer the data in as many
>> formats as possible but our most used protocol is WMS for
>> visualisation as overlays on maps. A the time (around 2011) we had to
>> patch thredds because we use relatively large (64gb)
>> netCDF-4 source files which are too large to load into memory using
>> the default BOUNDING_BOX reading strategy. Xiangtan Lin modified the
>> getOptimumDataReadingStrategy method in ncwms to make NetCDF-4 type
>> datasets default to SCANLINE reading strategy. This reduced the
>> memory footprint sufficiently to get the wms layers working. See discussion
>> here:
>> http://www.unidata.ucar.edu/mailing_lists/archives/thredds/2011/msg003
>> 28.html
>>
>>
>>
>> We are now consolidating this thredds instance into a large corporate
>> system and were hoping we could get this change included in the master
>> source so we can keep up with the current releases without having patch each
>> new one.
>> Comments in the source seem to suggest that SCANLINE is the intended
>> default for NetCDF files but the existing check only catches versions prior
>> to 4.
>>
>>
>>
>> So, would it be possible to have this change made to the ncwms master?
>> I can make the changes and setup a GitHub pull request if that helps.
>>
>>
>>
>> Thanks
>>
>> Peter Warren
>>
>>
>> _______________________________________________
>> NOTE: All exchanges posted to Unidata maintained email lists are
>> recorded in the Unidata inquiry tracking system and made publicly
>> available through the web. Users who post to any of the lists we
>> maintain are reminded to remove any personal information that they do
>> not want to be made public.
>>
>>
>> thredds mailing list
>> thredds@xxxxxxxxxxxxxxxx
>> For list information or to unsubscribe, visit:
>> http://www.unidata.ucar.edu/mailing_lists/
>
>
>
> --
> Dr. Richard P. Signell (508) 457-2229
> USGS, 384 Woods Hole Rd.
> Woods Hole, MA 02543-1598
--
Dr. Richard P. Signell (508) 457-2229
USGS, 384 Woods Hole Rd.
Woods Hole, MA 02543-1598
