[Thread Prev][Thread Next][Index]

Re: [ferret_users] longer input line buffer?



Thanks Andrew,

it is even better. Following the manual and defining a ferret-variable 'file_list' (using wildcards) and an aggregated virtual data set
works now. Great!
I did not read enough in the manual and overlooked the example to put the file list in a ferret variable.
So I went poor mans way and generated the file list by a shell script to produce a short ferret script that
defines the aggregated data set with a lengthy explicit file list. This is limited by the line buffer length. Having the information on the file names
in a ferret-variable, circumvents the line buffer limit elegantly.

Many thanks for the suggestion, it helped a lot.
Martin



On 08/16/2016 04:25 AM, Andrew Wittenberg - NOAA Federal wrote:
Filesystem wildcards (*, ?, curly braces) work in the aggregation command, so if you can arrange your files in that manner (e.g. using symlinks), this is a great workaround.  Note that Ferret automatically sorts files by the date coordinates, so you don't even have to do that.

Andrew

On Monday, August 15, 2016, Martin Schmidt <martin.schmidt@xxxxxxxxxxxxxxxxx> wrote:
> Hi,
>
> I am trying to generate time aggregations for numerical ocean model results. Usually, those data come
> in time slices. File names mark the period and may have 15 to 20 characters each. If the model results cover a longer period, the
> file list may become long and the aggregation command becomes longer than 2048 characters. In this case the command
> cannot be executed by ferret.
>
> If I would have a free wish ... would it be eventually possible to increase the input buffer size in future releases?
>
> With short file names the aggregation feature works perfect - much better than with the old descriptor files.
> Especially at compute centers the tredds-option is not permitted and aggregation with ferret allows a quicklook.
>
> I am using pyferret v7 compiled by myself. May be a hint would help, in which file the input buffer size is defined.
> So I could recompile and test by myself.
>
> Best,
> Martin Schmidt
>

--
---
Dr. Andrew T. Wittenberg
Physical Scientist (NOAA Federal)
Climate Change, Variability, and Prediction Group
NOAA/Geophysical Fluid Dynamics Laboratory
201 Forrestal Road
Princeton, NJ 08540-6649
USA
Tel: +1-609-987-5064
Fax: +1-609-987-5063
Web: http://www.gfdl.noaa.gov/~atw


[Thread Prev][Thread Next][Index]
Contact Us
Dept of Commerce / NOAA / OAR / PMEL / Ferret

Privacy Policy | Disclaimer | Accessibility Statement