Maarten Breddels

Depends on #1745 Needs some polishing and we now depend on ipyvuetify, which we shouldn't. Maybe put this in vaex-jupyter? 

when VAEX_PROGRESS_EXTRA=1 We see something extra:

✨ delete hdf5 file when export/convert fails

1

Otherwise we leave an inconsistent file that may lead other processes to think a file was property generated. This can happen when someone ctrl-c.

TODO: do the same for strings

This should speed up computations, since we don't go in and out of numpy. We also use the binner more consistently throughout. Also, this should allow for GrouperLimited, and Binner...

Should replace #1476 which makes Vaex more compatible with Pandas. cc @AlenkaF

✨ arrow/pandas cannot do all data conversion yet

1

e.g. dict encoded uint8/uint16 TODO: test uint16/32/64

✨ Lazy image reading support (reading, converting, resizing)

8

Example:  cc @xdssio

this serves as a reminder to open an issue at Arrow JIRA for 2 bugs in trim, see also #1186