ST: Finalizing the subtensor
Implementation.
#4
kannavkm
started this conversation in
Project Subtensor
Replies: 1 comment 1 reply
-
I favor your second proposal.
I am not sure if we need a change. Perhaps you could provide a small example how a new
Correct. However, |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Before beginning to work on the
subtensor
implementation, I wanted to get everyone's views on implementing it. To me, there seem, two candidates, as I highlighted in my project proposal as well. Note I am comparing them from a design point of view only, as it seems there would not be any performance differences between the two cases.subtensor
as an independent type: This is the current implementation present here. It would be easier to build on top of this.subtensor
as an engine. This would create a better design IMO since there would be nothing assubtensor
, and it would essentially be atensor
. However, this might not be as straightforward to implement as having a separate type as we might need some changes in thetensor
itself to accommodate a new engine.The
subtensor
is so that any library user would(normally) not create and only be created by member functions and operators oftensor
.Since we already have an implementation corresponding to
1
, maybe I could create an implementation on the same lines as2
. Maybe that could help in the decision-making process?Well, that's all that I think. But, of course, it would be great to see other's points of view as well.
Beta Was this translation helpful? Give feedback.
All reactions