Closed gtauzin closed 4 years ago
The tests are failing as, for two example files, there is a discrepency between the cell counts as outputed by flagser
and flagser-count
. As this is a problem I experience running C++ flagser
as well, I have opened an issue on the flagser
GitHub.
@ulupo My last commit adds a reshape to the dgms outputed by flagser_weighted
. It is necessary for FlagserPersistence
to deal with empty subdiagrams. It is also consistent with what we documented it should return, so I put it here.
@gtauzin thanks for the explanation, I was just about to ask about it!
@gtauzin thanks for the reminder. I'll find time for this over the weekend.
@gtauzin just reiterating the general question in my last review ("why doesn't flagser_weighted
get the coefficient treatment in a similar way as flagser_unweighted
?") in case it slipped.
@gtauzin just reiterating the general question in my last review ("why doesn't
flagser_weighted
get the coefficient treatment in a similar way asflagser_unweighted
?") in case it slipped.
I missed it, sorry. It's fixed as there is no reason, I just forgot.
General question: why doesn't
flagser_weighted
get the coefficient treatment in a similar way asflagser_unweighted
?Generally LGTM, just left a couple of comments. The C++/pybind11 part is slightly above my pay grade, I wonder if @MonkeyBreaker could give it a quick look?
Hi, sorry for my late reply, I can give a review this afternoon if it's fine for both of you :)
@MonkeyBreaker yes please :)
LGTM !
Changes are good for me :)
What does this implement/fix? Explain your changes.
Due to high demand,
pyflagser
is enabling the direct use ofcount_cells
functionalities from C++flagser
.Additionally:
43 is fixed
Any other comments?
This adds the pybind11 bindings, the python API and the appropriate tests.