At AskAnna we release new features, changes and fixes once they are ready to deploy. On the Beta of AskAnna we always run the latest approved code, also if it does not have an official version number yet. From time to time we create a new version. In the release notes you will find AskAnna versions and everything we release to Beta that is not part of an official version. These non-version releases will be listed here with the date of deployment to Beta.
In these release notes we give updates about the AskAnna platform. Besides the platform, we also have the AskAnna Python project that contains the AskAnna CLI and Python SDK. Release notes for this project are available in the AskAnna Python changelog published on GitLab.com.
We now work on making AskAnna an open-source project. And on extending the AskAnna CLI with an option to run and track your local runs. If you have ideas for tracking code, data and experiments, please send us an email!
- Viewing Jupyter Notebooks
- Running the job in the web interface now by default doesn't show your token anymore. We added a hide/show option for the token.
- Running the job: new examples with how to run using the AskAnna Python SDK & CLI
- CLI: get more info about your runs in the command line using the new commands
askanna run statusand
askanna run log
- Python SDK:
- Run SDK: it's now also possible to get the run result, artifact, payload and log
- Run SDK & Job SDK: added delete and change a run or job
- Workspace SDK & Project SDK: added create and delete workspaces or projects = Result SDK: download a result to a provided path
- Added relation type data classes and updated existing data classes to use these relation type data classes
- The duplicate
short_uuidthat identify an object in the API response are now replaced by
- Cleanup AskAnna OpenAPI docs
- In the AskAnna Backend we split the JobRun app into a Job and Run app. With this change we also renamed some API endpoints:
- Starting a new job run is moved to
/v1/runinfo/endpoints are moved to
- Run info related endpoints
/v1/statusare now part of
- Starting a new job run is moved to
- In the API responses we now always show relation type object when a reference to another object is part of the response
- Refactor Python SDK: align setup of gateway methods & CLI commands
- Scroll on the run metric table view now uses more page height
- Copy the log now also works for active runs and will copy the last available log
- Sorting & filtering on run metrics and variables tables
- Support for GitPod
- A redirect to the original requested page in case someone needs to sign in
- In case created by or people name info is missing from the API response, we display a fallback for the name
- Refactor app to use Vite
- Downloading code archives
- Removed unused API endpoints for project templates
- Updated API documentation to OpenAPI Specifictation v3.1.0
- Added support to authorize requests with a token in the API documentation
- Aligned run status labels to always display the same value format
- Refactoring of frontend to use Pinia, change folder structure and remove unused components
askanna pushnow also works when you are signed in to another URL then
askanna create --pushnow really pushes the initial version of the code after creating a new local project
- Compare runs: an initial version of a page where you can compare the last 15 runs directly in AskAnna
- In the API the JSON response you get from the runinfo-endpoint is changed. Now the response contains more metadata. The metadata is used, among other things, for the new compare runs page.
- In the API the authentication URLs were not versioned. In this release, we changed this. The AskAnna Frontend and Python SDK/CLI are also updated to support this change.
- CLI: when you are already logged in and try to log in again, you will get the question if you want to log in with another account. It's not necessary anymore to first log out.
- Deduplicate of run metrics and variables did not work anymore after a recent update of Django. This release fixes the function to deduplicate run metrics and variables, including a fix for historical runs.
- The examples of running a job in the AskAnna Frontend show your token again.
- Rendering Markdown files
- Autoloading of "many" records for run metrics table and run variables table
- Visualize metrics: it's now possible to make a graph from your metrics
- If a metric or variable is of type
list, we try to detect the type of data in the list and set a more specific type of data to the metric record. See also the metrics documentation.
- Python variables of type Numpy array can now be stored as a metric/variable.
- In the run JSON response besides the unique metrics & variables names, we also added the type of data.
- Calculation of table height for run metrics & variables, so that also the last row is fully visible.
- Duration of runs less than 1 second resulted in a misrepresentation of duration. If a run lasted less than 1 second, you now get the value 0 seconds.
- "Hamburger"-menu works again
- The API endpoint to get the metrics and tracked_variables
metais removed. This information was also available in the
metaendpoint was not used and with the changes in this release, we decided to remove the endpoint.
- Create workspaces: it was only possible to create a workspace when you created a new account. Now, you can also create a new workspace when you are signed in.
- Search, sort and filter your workspace projects.
- On list pages there is a new filter icon. The icon is filled when a filter or search is active on the presented list.
- Search, sort and filter options are no longer visible when you open the explore workspace and project pages. Click on to get the search, sort and filter options. If a filter, is active the icon is filled .
- On the workspace list the options on the workspace card now also contain edit & remove workspace. You only see these options when you have permission to perform them.
- Response of workspace API endpoints changed. The response now also contains the user who created the workspace, the permissions of the authenticated user on the workspace, and the URL of the workspace.
- Switch to Backup Helper for making database and file backups
- CSV and TSV files now also have a pretty view. When you open this type of files we will try to show you a table with the file's content.
- Excel files now have a pretty view. When you open this type of files we will try to show you the Excel sheets and per sheet a section with a table with the content.
- Only register a run as succeeded if the last line of the log contains the success message.
- When a run fails, add a message
Run failedto the log.
- Deleted background from code sections.
- When switching projects, always load data for the latest opened project.
- On the code version, open the specific historic version of the file.
- The job name on the job page is back.
- Markdown views now open by default marked up, or as we call it in a pretty view. You can also switch to the raw version of the file.
- If your code directory has a
README.mdfile, we show the marked-up README below the list of files and directories.
- On files that contain a pretty view option, you can now switch between pretty and raw views via a button in the file bar instead of a drop-down field.
- Added view menu to HTML files to make it possible to see the raw HTML file.
- CLI: Added an argument 'url' to
askanna login, making it possible to log in on different AskAnna environments using the web URL
- Refactor for file handling. Code, input, result and artifact files are now handled the same way.
- Improve last run status calculation. It's now based on the latest status time i.s.o. the creation time of the run.
- Python SDK/CLI: make the requirements less strict, so it's easier to integrate AskAnna in your existing Python environment.
- On the run log, the loading dots will now disappear after loading the entire log.
- When opening AskAnna for the first time, we will not show the message that a new version is available. We have already loaded the latest version for you.
- Fixed opening Jupyter Notebook files
- Make new projects directly available in the project menu after creation.
- Update breadcrumbs after changing project or job name.
- CLI: after successful login, you now get a success message i.s.o. a fail message
- Upgrade of editor with:
- support for links
- fullscreen mode
- description fields now have more space
- improvements to get out of a code block and continue with normal text
- a fix where empty lines got removed
- another fix where we added a new line after leaving a code block using the ESC key
- Copy buttons for run result and environment image name
- Added support for copy image on code and artifact view
- Added decoder for breadcrumbs title, so we don't show spaces as
- Backend now returns a JSON with detail about invalid token
- Code highlight of run results instead of showing plain text
- The description editor got a sticky header so you always see the markup icons
- The run metrics table now has a dynamic height
- Status colors are aligned between notifications emails & frontend
- When a run status is queued or pending, we don't show a duration of 0 anymore, but we inform you that the run is
- Look-and-feel for code package instructions is aligned
- Instead of calculating the duration in the frontend, we always use the duration provided by the backend
- Changed text for non-existing file or directory on code view
- Fixed the title in the edit form for job description
- Fixed loading run logs that are longer than 200 lines