This releases is a new major release with some new features and some backwards incompatible changes that require a careful upgrade and some manual intervention. The 0.7.x releases will also be that last to support Python 2. If you haven’t upgraded to Python 3 yet, we advise you to do now.
BC break The major change in this version is no longer initiating the skosprovider.registry.Registry on starting the application, but when a request is created. The previous way of working created problems with SQLAlchemy providers in a webserver using mutiple threads. Please review the docs at https://atramhasis.readthedocs.io/en/latest/customisation.html#creating-conceptschemes to see how it works now. For more background, have a look at the pyramid_skosprovider docs at https://pyramid-skosprovider.readthedocs.io/en/0.9.0/install.html
- All requirements were updated to their latest versions. Python versions were fixed to 2.7, 3.6, 3.7 and 3.8. If you made custom changes, you might have to edit them. (#508, #519, #513, #566)
- Npm has replaced bower as the package manager for frontend packages and the build process was revised. If you made custom frontend changes, please check them thoroughly.(#511)
- Instantiation of the SKOS registry was changed to work on a per request basis. (#346, #490, #535)
- Fixed a major issue with generating the expanded version of a concept. By default the assumption was that concepts in a collection were also narrower concepts of the collection’s superordinate concept, but the implementation for this was incomplete and contained bugs. This has been changed to an boolean attribute infer_concept_relations. When set to true, concepts in a collection are considerd to be narrower concepts of that collections’s superordinate concept. This is especially important for a provider’s expand function and affects what is considered a narrower concept of a concept that uses thesaurus arrays or node labels.
- The docs were updated and now contain a part detailing what Atramhasis does with some screenshots. (#495, #583, #440)
- Default inclusion of skosprovider_heritagedata was removed because the service is unstable too often. (#537)
- Improve some SEO by adding canonicul URL’s, open graph info, Twitter cards and the ability to generate a sitemap through a script. (#530, #531, #496, #497)
- Clean up importing and exporting of conceptschemes to make it easier. (#452, #475, #476, #495)
- Provide a simple, printable version of a thesaurus tree. (#533, #532)
- Add a script to make removing a conceptscheme easier. Be careful as this will drop all concepts, collections and the conceptscheme itself. (#579)
- Lots of minor improvements and bug fixes.
- Fix corrupt build
- Security updates
- Update Colander and other dependencies. (#464)
- Remove old convert_oe script because it depends on an obsolete webservice. (#466)
- Fix an issue with circular dependencies in requirements files. (#463)
- Change the default GA macro to anonymizeIp and be more GDPR compliant. (#450)
- Generate a default dump location in development.ini files. (#416)
- Update skosprovider_sqlalchemy to solve a problem with the tree cache. (#455)
- Update a lot of dependencies.
0.6.3 was a brown bag release. Sorry!
This version is exactly the same as 0.6.3, but properly packaged.
- This version updates a lot of the requirements to their latest versions. This might be an issued if you’ve written lots of code against older pyramid versions. (#418, #413, #412, #411, #410, #408, #407, #404, #403)
- Remove authentication from the demo version since Persona does not work anymore. (#361)
- Fixed the base HTML template and added a DOCTYPE declaration. (#429)
- Update the URI’s for licenses of the Flemish Government. (#430)
- Fixed an issues with sorting on Python 3. (#424)
- When an LDF server is present, add a link to the HTML document to this server. (#394)
- Wrong expansion of SKOS namespace in LDF server. (#401)
This release is a minor release, containing improvements regarding the Linked Data Fragments server.
- Also add hidden datasets to the LDF server. Only external ones are not added now. (#390)
- Make it possible to set the LDF server protocol when generating the config. (#391)
- When generating an LDF server config, add a composite source as well. (#393)
- When generating the dataset information, add hydra controls that link to the LDF server instance. (#392)
This release is a major release containing new features.
- Added a script to generate nightly dumps. Instead of generating full downloads on demand, they can now be generated by a cron job (eg. once per night, week, …). This makes it possible to download a large conceptscheme at once. During these dumps, some statistics on every conceptscheme such as the number of triples in it will be generated as well. This was done to make it easier to embed a custom Python based LDF server, but currently only serves the purpose of keeping some score. (#337, #360)
- Added easy integration with a Linked Data Fragments server (https://linkeddatafragments.org). Atramhasis can now generate a config file for such a server that you can use to setup the server. By default this config will work with the Turtle files that can be generated every night. But if you have access to the HDT library, you can also work with HDT files for a masssive performance boost. See the section Running a Linked Data Fragments server in the docs for more information. (#365)
- Add some more information the HTML title tags for a concept detail. (#363)
- Changed the UI for doing a search so that you now get a proper warning when searching for a label without specifying the conceptscheme to search in. (#373)
- It is now possible to generate URI’s when importing from a file that does not contain them, eg. a JSON or CSV file. The import_file can now take a pattern_uri parameter than will be used to generate new URI’s with. (#372)
- Fixed some issues with the tree cache that came to light when running Atramhasis as two nodes. Where before it was not possible to configure the tree cache, it now is. Previously an in-memory cache would always be used. Now it’s possible to use a different type of cache. If you’re running more than one webserver, it is advised to run a shared cache. If you’re running a previous version of Atramhasis, you will need to configure your cache with cache.tree and cache.list settings. (#371)
- It is now possible to add sortLabels to concepts. These can be used in the REST service to arbitrarily sort concepts. The sortLabel works per language. This makes it possible to eg. sort historical periods in chronological order. Most of the functionality was already present in skosprovider 0.6.0, but it had not been properly included in Atramhasis. (#362)
- Added ‘und - undetermined’ to the default language set to support json file imports. (#386)
- Fixed a bug when editing concepts where data from previously opened concepts would bleed into the concept you were editing. (#367)
- Update several dependencies to the latest versions. (#380, #381, #376)
- Added ‘und - undetermined’ to the default language set to support json file imports (#386)
This minor release fixes a bug with the protected resource event. The event should give the uri of a concept instead of the url path. In addition to the uri the request is added to the event. It also fixes the bug with removing relations and updates the requirements for skosprovider_sqlalchemy.
This minor release fixes a bug with the tree browser. Before it wasn’t possible to zoom and pan the tree. With certain larger trees this would cause issues as content would run off the page.
This release is a major update based on the skosprovider 0.6.0 line of libraries. The most visible change is with the public and admin interfaces. These have been completely overhauled to provide a more pleasing user experience. Among other things visitors are now pointed towards popular concepts and concepts they have recently visited. Browsing an entire conceptscheme tree has been redesigned.
The adming interface now offers users an option to edit certain aspects of a conceptscheme such as the labels, notes and sources. Editing in general has been update and improved. Links between the public interface and the admin interface have been added to make switching from one to the other easier. Notes and sources can now contain certain HTML tags, allowing greater flexibility in defining concepts and collections.
A command line script was added to make it easy to import an entire conceptscheme, eg. when migrating from another system. It is now possible to import a RDF, CSV or JSON file on the command line in your Atramhasis instance. With earlier versions you had to script this yourself.
As always, bugs have been fixed, code has been rewritten and documenation has been updated.
See https://github.com/OnroerendErfgoed/atramhasis/milestone/8?closed=1 for the full list of changes.
- Added more sample datasets to get a better view of real data. These will make the demo more interesting.
- Fix a bug where it was possible to create a relation between a concept and itself causing all sorts of nasty things to happen.
- Minor refactoring. Move the pyramid routes to a new file.
- Added a CONTRIBUTING.md file. Contributions welcome!
We had some packaging issues with the 0.4.2 release.
This release of Atramhasis is mostly a bugfix update of the 0.4.1 release.
- Fix paths of db in scaffolds
- Add more information on exceptions
- Update skosprovider_getty and skosprovider_heritagedata (fix the problems when importing external thesauri)
- Documentation update
This release of Atramhasis is a minor update of the 0.4.0 release, focussing on small corrections and improvements and improving the documentation. A few interesting non-invasive features were added, mostly to the editor’s admin interface and machine-readable exports of RDF data.
Upgrading from 0.4.0 should be simple and cause no or few problems.
- A conceptscheme, concept or collection can now be exported to RDF through
skosprovider_rdf 0.3.1. These are individuals export endpoints that can
be reached in one of two ways. Either by hitting a url like
http://localhost:6543/conceptschemes/GEOGRAPHY/c/335 with a supported RDF mimetype
text-turle). Or by using an RDF syntax specific suffix (.rdf or .ttl).
- When importing, allow the user to request more information on a concept or collection, before actually importing it.
- Allow merging a concept with other concepts it matches. This allows a user to compare a local concept with an external one it matches and import any notes or labels that are present in the external concept, but not the local one.
- Reworked some parts of the public interface to make everything a bit clearer and to make all pages easily reachable.
- Allow sorting the languages in the admin interface.
- Reorganised and extended the right click menu on the grid in the admin interface.
- Allow looking up a skos:match from within the admin interface.
- Some issues with the length of language ids were solved.
- Fixed some issues when importing a collection instead of a concept.
- Made it easy to add a Google Analytics tracker.
- Added instructions on how to deploy a demo site on heroku. These work just as well for deploying an actual production site to heroku.
- Lots of small updates and tweaks to the documentation.
- Updated some dependencies.
- Some code cleanup and reorganisation. Several smaller bugs in the admin interface were fixed.
- The data fixtures were updated with skos:note examples. Added a license for reuse of the fixture data.
- Update to skosprovider 0.5.0. Among other things, this makes it possible to handle relations between Concepts and Collections using the subordinate_arrays and superordinates properties. Conceptschemes are now also much better integrated within the providers, thus making it possible to provider more context for a Concept. This version of skosprovider can also handle skos:matches.
- Add possibility to edit language tags. It’s now possible to use the admin interface to add, edit and delete languages in Atramhasis.
- When the REST service receives labels or notes in currently unavailable languages, it will validate those through language_tags. It the languages are valid according to the IANA registry, they will be added to the languages available in the application.
- Default length of language id changed to 64 characters. This is not available as an alembic migration. So only effective when creating a new database. If you already have a database created from an older version of Atramhasis, please modify accordingly. Modifying column length on SQLite is not possible (see http://www.sqlite.org/omitted.html ).
- Abiltity to match Concepts in an Atramhasis ConceptScheme to Concepts in external ConceptSchemes through properties such as skos:exactMatch and skos:closeMatch.
- Ability to import Concepts and Collections from external providers. This makes it possible to import Concepts from eg. the AAT (via skosprovider_getty), Flanders Heritage Thesauri (via skosprovider_oe), English Heritage Thesauri (via skosprovider_heritagedata) or any other SKOS vocabulary for which a skosprovider has been written. Currently only the concept or collection itself can be imported, without its relations to other concepts or collections.
- Add the ability to have a delete of a concept or collection fail if it is being used in other systems.
- Implement a delete permission.
- Add validation rule that a Concept must have at least one label.
- Update to skosprovider_sqlalchemy 0.4.1.
- Update to pyramid_skosprovider 0.5.0.
- Update to skosprovider_rdf 0.3.0. This update adds support for dumping ConceptScheme in an RDF file and also handles subordinate_arrays and superordinates.
- Update to language_tags 0.3.0.
- Update to skosprovider_sqlalchemy 0.2.1.
- Update to skosprovider_rdf 0.1.3 This fixes an issue with RDF having some SKOS elements in the wrong namespace. Also added a missing dependency on skosprovider_rdf to setup.py
- Updated the Travis build file to run a basic dojo build and test for build failures.
- Atramhasis now includes a working admin userinterface at /admin. Still needs some polish when it comes to error handling and reporting about validation errors.
- The admin module gets run through a dojo build to minimize page loads and download times
- Added RDF/XML en RDF/Turtle downloads to the public interface. Currently only dumps a full conceptscheme, not individual concepts.
- Added more docs.
- Full public userinterface
- REST CRUD service
- Security integration
- CSV export
- demo using Mozilla Persona as sample security setup