https://codemirror.net/demo/activeline.html

since be31425e

not only complete python language words but also all words from the
buffer.

/reviewed-on nexedi/erp5!122

Mark the test as expected failure for now.

This test is in expectedFailure state and is still not passing with this
change, but this test was showing wrong example that can cause
confusion.

Here are 2 examples of fixed cases:

- File object with data=None was exported with a second file containing the
  string 'None'

- Zope Pages Templates (and derived OOoTemplate) relied __setstate__ to fix
  the wrongly imported object.

This avoids computing a file extension that may not be used (if there's no data
to export), and fixes unreliable code duplication in _guessFilename().

Plus more implicit acquisition fixed.

We'll add another test checking that nothing is exported as a separate file.

Such images don't seem to exist.
On the other side, it may be useful to test raw data.

Note that the code to get data duplicates what's done later in export().
One should problably guess extension at the last moment.

Contrary to what the code looked like, it was always using /tmp.
Which could also cause conflict with other test suites running at the same time.

This was breaking the next test.

We should have done before committing 76ecef89
but doing it now with a dummy merge also works.

Unfortunately, this is not really useful for blame, because all lines are
rewritten, at least to remove the trailing '\n'.

Files that have already been modified in 'master' branch are excluded.

Desactivate DMS to speed up restrictedTraverse document retrieval and so, speed up jIO sync

/proposed-by @jerome

Use MIME encoded words also for names.
https://en.wikipedia.org/wiki/Email#Message_header

It would be surely better to have matrixbox having possibility to
define which widget we use to display matrix variation labels.