Merge pull request #10562 from ipython/auto-backport-of-pr-10558 Backport PR #10558 on branch 5.x

Backport PR #10558: Let nbformat take care of opening the file Notebook files should always be read as utf-8, but the default for `io.open()` is to use a platform-dependent default encoding. The easiest way around this is to pass the filename to nbformat.read() and let it open the file correctly. Bug identified at: http://stackoverflow.com/questions/43915006/encoding-error-in-jupyter-when-run-another-notebook

Merge pull request #10557 from ipython/auto-backport-of-pr-10543 Backport PR #10543 on branch 5.x

Backport PR #10543: Pass down include and exclude as kwarg to repr_mimebundle Followup on 10526

Merge pull request #10550 from ipython/auto-backport-of-pr-10539 Backport PR #10539 on branch 5.x

Backport PR #10539: Respect the ignore and include keyword in display Fixes 10526 , but as noted in there I think we should think about passing extra arguments to `_repr_mimetype_` I even think we __should__ require that it takes arbitrary kwargs to be able to pass more informations later on without breaking API. cc minrk and rgbkrk

Merge pull request #10544 from ipython/auto-backport-of-pr-10496 Backport PR #10496 on branch 5.x

Backport PR #10496: Define `_repr_mimebundle_` Allows objects to display arbitrary mime-types by returning a mimebundle. This is getting increasingly important as custom mime-types are growing in popularity. - mime-bundle is computed first, but other formatters are still called - if a mime-type is present in repr-mimebundle, `_repr_<mime>_` will not be called (avoids redundant calls for backward-compatible objects) closes 10090 cc rgbkrk Alternative design: rather than single method returning the mimebundle itself, return mime-keyed mapping to callables, e.g.: ```python def _repr_mime_methods_(self): return { 'text/html': self._repr_html_, } ``` Another more minor alternative: rather than allowing return of `data` or `(data, metadata)`, require returning the full mime-bundle with `data`, `metadata` keys: ```python def _repr_mimebundle(self): return { 'data': { 'application/vnd.foo+json': [1,2,3], }, } ```

Merge pull request #10505 from Carreau/backport-of-pr-10048 Backport PR #10048 on branch 5.x

Merge pull request #10524 from ipython/auto-backport-of-pr-10523 Backport PR #10523 on branch 5.x

Backport PR #10523: Don't reload __mp_main__ Fixes 10522.

Fake the Python 3 keyword only arguments.

Backport PR #10048 on branch 5.x Implement update_display (94d6f5bc677f8ed924f46d79badaa5777b4820d5)

Merge pull request #10504 from ipython/auto-backport-of-pr-10414 Backport PR #10414 on branch 5.x

Backport PR #10414: update author email now that list is on python.org

Merge pull request #10501 from ipython/auto-backport-of-pr-10489 Backport PR #10489 on branch 5.x

Backport PR #10489: Prefer execution when there's only a single line entered Closes gh-10425 The heuristic here is to treat a single line specially, and always evaluate it as if the cursor was at the end. An alternative heuristic could be to do this if the cursor is on the last line of the input. This could also cause some weird effects if you e.g. type `for a in range(5):`, move the cursor back a few places and press enter - you'll get a newline inserted in the text, but it will indent as if it were after the colon. I'm still trying to think if there's a better way to approach it.

Merge pull request #10481 from ipython/auto-backport-of-pr-10479 Backport PR #10479 on branch 5.x

Backport PR #10479: Make pdb use pt style. Fixes #10478

Merge pull request #10476 from frenzymadness/5.x Fix bug in generating shortcuts