Improve mimetype detection for remote files that
have no file extension. The mimetype detection
that makes use of the file, only works with local
files, so do not use the remote path which can be
a URL or stream.
- Method names renamed to avoid interface changes once
we support regression and unsupervised learning
- Adding regressor interface even if not implemente
- predictor interface comments expanded
- Differentiate model's required accuracy from predictions quality
- Add missing get_callback_boundary call
- Updated datasets' metadata to allow 3rd parties to code
regressors themselves
- Add missing option to exception message
- Include target data into the dataset regardless of being a prediction
dataset or a training dataset
- Explicit in_array and array_search non-strict calls
- Overwrite discrete should_be_displayed implementation with the binary one
- Overwrite no_teacher get_display_value as it would otherwise look
wrong
- Other minor fixes
We want to avoid page jumps wherever possible. This change uses the original select element as a "proxy" to reserve the space
in the DOM for the enhanced auto-complete when the JS runs.
It uses visibility: hidden to make the select not rendered - but still take up space.
The exact sizing for the CSS was determined by testing and reverse engineering the bootstrap calculations.
The size of user pictures in selectors was reduced to avoid flicker when switching between no selection and a selected user. I
could have reserved a larger amount of space, but it looks worse for pickers with no pictures.
Instead of putting vertical-align: middle as indication for participants table
headers, none is set which leads to theme default, i.e. bottom in Clean and Boost.
Before the patch there were some problems with sorting
causing student1@ and student10@ to be sorted inconsistently
between databases.
Where that is not an important problem in real use, was
causing tests to fail consistently in sqlsrv.
So, adding student0@ to the tests in order to make them
stable collation-wise.
