Opened 2 years ago

Closed 2 years ago

#2231 closed help (answered)

Have been asked to build a different version of XConv

Reported by: markr Owned by: jeff
Component: UM Tools Keywords: xconv, source, linux
Cc: Platform: Other
UM Version: 4.5

Description

Hello,
I wonder if it is possible to get the source for XConv utilities.
The palaeo group at Leeds are still using vn4.x and have used a bespoke version to work with the PP files. They have tried to use the xconv-1.93 provided as a pre-built exec but it does not behave as anticipated. Their previous version fails to run in the Centos7 environment.

Can you help?

Many thanks,
Mark

Change History (4)

comment:1 Changed 2 years ago by jeff

  • Owner changed from um_support to jeff
  • Status changed from new to accepted

Hi Mark

What version of xconv are you currently using? What is the problem with xconv1.93? The xconv source code isn't generally available, what are the changes you need to make?

Jeff.

comment:2 Changed 2 years ago by markr

Hello Jeff,

partially solved this one because the faculty IT support had a copy of 1.91 that had been installed on Polaris (n8hpc). We were able to build on ARC3. [I had already installed your 1.93 package but they observed as quoted below]

The Palaeo group at Leeds School of Earth and Environment have reasons for sticking with version 4.5 of the UM. Also they are particular about the version of xconv/convsh and um2nc; they seem to prefer vn1.91.

I think there are several issues and I better to pass on their observations:

Here is what their PI told me:

Essentially, we use convsh to convert from pp file format to nc
format. When we do this we convert some of the not very helpful
variable names to something with more meaning. This helps prevent
errors during analysis. For this, there are look-up tables that my
version of um2nc uses, but I think xconv (and/or convsh) may also need
to be recompiled for this to work because when I tried it with your
version 1.93 it didn't use the names in the tables but reverted to the
standard names even though um2nc was pointing to them.

An additional issue I was shown is that when viewing a PP time series it showed strange behaviour displaying interval of a different period (I think it was supposed to average over 500 weeks) showing 504 weeks (or 540 I forget).
Also there was a display of negative time after 8 slices had been read.

I told them that after I had installed on ARC3 then if these "errors" persist then they should raise a ticket with CMS NCAS as it really is not my area of expertise.

I hope that is a sensible option that they should liaise directly with yourselves… but initially i did not know where to get the 1.91 to build on ARC3.

Thanks for the follow up. Should we close this now and wait for them to decide to follow up?

Mark

comment:3 Changed 2 years ago by jeff

Hi Mark

I do remember there was a modified version of xconv in use by some group which used lookup tables to modify the names, I guess this is that version. The current version of convsh allows you to modify the names in the script so maybe this could replace the older modified version, although this won't work for xconv.

AS for the other problem, yes it would be best to raise another ticket for that. Although xconv is no longer being developed except for fixing bugs, time permitting.

Jeff.

comment:4 Changed 2 years ago by jeff

  • Resolution set to answered
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.