v0.0.52 RC Release
As of yesterday v0.0.52-RC2 has been released and is running on data.gov.au, and 0.0.51-RC5 has been promoted to 0.0.51!
Here are some highlights of 0.0.52:
CKAN Interoperability
Data.gov.au has been using CKAN for a month now, and as a part of that we’ve discovered a few things we missed in terms of interoperability. For instances of Magda that are set up to redirect CKAN routes to a CKAN instance, a few routes that we missed will now redirect correctly — /dashboard and /dataset?q=keyword.
We’ve also removed the hard-coding of the CKAN sign on provider — now you can point it at any CKAN instance rather than being stuck with data.gov.au (so it supports the new location of DGA CKAN at data.gov.au/data. Sadly the UI still says data.gov.au, but if you need to make this customisable please send us a PR!
AAF Support
One of our developers Jacky has spent a bunch of time doing work for our friends over at CSIRO Environmental Informatics, who are running Magda as part of their Knowledge Network v2 project. Part of this was adding support for the Australian Access Federation as a sign on provider — as this has been merged up into the main Magda repository, now anyone who wants to provide SSO support to users from Australian scientific institutions can turn this on — it even works with the main Magda UI!
UI Improvements
There have been a few small UI improvements in this release.
Previously datasets with no landing page to link to ended up hyperlinked at the bottom of the dataset page anyway… the hyperlink just didn’t go anywhere (or more precisely, it linked to where you already were, because the tag had href=””). As of this release the UI now handles this properly by only displaying a link if it has something to link to:
Also after some consulting with our friends over at the DTA, we’ve changed what used to be called “Open Data Quality” to “Linked Data Rating”:
This more accurately reflects what the rating is actually about. Admittedly not everyone is familiar with what Linked Data is, but our hope is that this will prompt them to learn about it and advocate for more of it!
Behind the Scenes
One of our Data61 colleagues Mike completed an epic 1000-line refactor of the Registry API in order to improve its performance, which has been efficiently whirring along on our dev server for a few weeks and is being released as part of 0.0.52. Previously we’d see the Registry API quite often using half a CPU just to process webhooks, as well as occasionally completely failing to respond to liveness checks and sometimes sending multiple webhooks for a single event — this should resolve these issues to some extent.
Internal Magda
This year we’ve also deployed private instances of Magda for internal use at our partners in the Australian federal government — the Department of Agriculture and Water Resources and Department of Environment and Energy. Running Magda in this configuration is very different to running it as an open data portal, and in this release we’ve fixed a few awkward issues that doing so exposed.
When searching for datasets in Magda, the score of the results is influenced by their Linked Data Rating, and this influence is done by multiplying the score by their rating out of 1… however, what we found running Magda with no Linked Data Ratings is that when the data rating is 0, the score of all datasets returned is 0… so there’d be no ranking of results at all, just filtering. This release changes the scoring so that even datasets with 0 stars are ranked correctly relative to each other, but behind other, higher-rated datasets.
We’ve also made a minor improvement to the CSV connector — it now copies descriptions from datasets into distributions rather than displaying “No description” when clicking on files.
Migration
This release is a small one — no need to do anything manual, just upgrade with helm as usual.
How to get it
The new version is available as a helm chart at https://charts.magda.io, and you can try the new features out at https://data.gov.au. Full changelog is here.
