[LTER-im] ECC checks

[Margaret O'Brien]

HI all -
At the water cooler April 10, we introduced a group of new/redesigned
checks. It was requested that they go into production after a bit of time,
so that people had a chance to digest the affects (which are small). We
plan to install the checks to the production pasta system at the next
regular pasta update, which will be on the evening of *Wednesday, 2017 May
3*

I have fwd'd the message summarizing the checks, in case you forgot what
they were. the PDF slides are here: http://im.lternet.edu/node/1300

Remember - these new checks will only affect new or updated datasets; it
will NOT affect datasets already in pasta.

In a nutshell, the new checks are to ensure data/package integrity
numberOfRecordsPresence (warn)
alerts the submitter if the numberOfRecords element is not included. The
presence of the element is checked before the numberOfRecords check is run
(which compares asserted numberOfRecords to observed)

pastaDoiAbsent (error)
the system will not accept package which include a pasta DOI. PASTA assigns
the DOI during the generation of L1 EML.

integrityChecksumPresence (warn)
analogous to other presence checks, looks for an element before using its
content

integrityChecksum (error)
The system will reject packages where the asserted integrity checksum does
not match a computed checksum at upload.



In other news -- the ECC WG has reformed, and is meeting regularly. Our
first task is a check on dateTime formats. This will include
recommendations to the EML schema and its documentation, for eml-2.2.

best,
Margaret
----


Margaret O'Brien
Information Management
Marine Science Institute, UCSB
Santa Barbara, CA 93106805-893-2071 <(805)%20893-2071>
(voice)http://environmentaldatainitiative.orghttp://sbc.marinebon.orghttp://sbc.lternet.edu


---------- Forwarded message ----------
From: Margaret O'Brien <margaret.obrien at ucsb.edu>
Date: Mon, Apr 3, 2017 at 4:50 PM
Subject: Monday's Watercooler, ECC checks
To: all_LTER_IMs <im at lternet.edu>


Hi all -

As you know, an LTER IMC watercooler is scheduled for next Monday, April 10
(3pm EDT). One of the topics will be advances planned for the ECC - the
system performing dataset checking for PASTA. This msg is a short
description of what we would like to cover.

In case you have forgotten, back in 2012 an IMC working group finalized 72
checks, and ~25 of these were running when PASTA went into production in
2013. In the intervening time, additional checks were implemented depending
on resources available, and other checks proposed.  Fast forward to today:
EDI is up and running, and we have resources budgeted to work on this more
systematically.

The new checks to be implemented are related to specific feature requests.
These involve data integrity, and they are important to review with you
because failure will generate an 'error' and block upload of the dataset.

1. checksum (2 checks, details on request):
These will confirm entity integrity during upload. The checksum can be used
later by PASTA to minimize entity duplication.

2. DOIs:
PASTA now adds package DOIs to L1 EML. This means that L0 EML should not
contain a DOI (e.g., a DOI may have been inadvertently left behind if an
EML doc was recycled).  This check will prevent confusion due to possible
conflicting ids.

During the watercooler, we will outline specifics about the checks. As with
other PASTA improvements, checks will be developed and can be tested on
portal-d (i.e., you can pre-evaluate your trial EML), and portal-s is
reserved as the staging platform for production. For a summary of the
checker, its behavior, and results from the first few years with LTER
datasets, see this paper in the recent Ecoinformatics special issue, DOI:
10.1016/j.ecoinf.2016.08.001

Best,
EDI team




-- 
-----------
Margaret O'Brien
Information Management
Marine Science Institute, UCSB
Santa Barbara, CA 93106
805-893-2071 (voice)
http://environmentaldatainitiative.org
http://sbc.marinebon.org
http://sbc.lternet.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lternet.edu/pipermail/im/attachments/20170428/0627c5dd/attachment.html>