-
Notifications
You must be signed in to change notification settings - Fork 74
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Resolve sparse field in Spec #139
Comments
is this disabled because of this issue, or the conversion?
If we remove the sparse bool and implement a NTA_BasicType_Sparse:
Isn't that too much extra work&overhead, just compared to the isSparse field? And an argument for keeping the status quo (with sparse bool):
the Hence, my vote to keep as is. Or at least keep now and might change if someone has objections. See if we can run #137 |
the isSparse does not come from our community-era changes, it's in
Numenta master.
Oh, that complicates things.
If we don't use NTA_BasicType to handle different data types or format of
arrays, then we don't have a means to handle the SDR object either. There
is no way for the framework to pass an SDR from one region to another.
…On Mon, Dec 3, 2018 at 6:00 AM breznak ***@***.***> wrote:
So for now I am going to disable some of the LinkTests so I can finish
this PR, but we need to go back and fix is.
is this disabled because of this issue, or the conversion?
Now...the current implementation does not allow array data type
conversions within the Links between Regions. It will require that the
Output type of the sending Region must have the SAME type as the receiving
Region's Input. One of the future PR's will be to provide some portion of
automatic data type conversion within the Link.
If we remove the sparse bool and implement a NTA_BasicType_Sparse:
So by wrapping your new buffer types in an Array you can pass your sparse
or SPR buffers through the framework to the various Regions.
Isn't that too much extra work&overhead, just compared to the isSparse
field?
And an argument for keeping the status quo (with sparse bool):
The OPF framework mentioned in the API (all Python code) relies on this
facility in the Network framework in order to do its job so we have to be
careful that we don't break it.
If you want me to do those changes I can. But creating a new field in the
Spec is not the way to do it. Lets use the architecture as it was intended.
the isSparse does not come from our community-era changes, it's in
Numenta master. Meaning the API doc is out-of-date and correcting/removing
sparse would actually break the py code and any applications currently
using it.
Hence, my vote to keep as is.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#139 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AFBa_6bPpTxxu6SQRZO4o4gpz5bbfz1tks5u1S55gaJpZM4Y-nyW>
.
|
we can use NTA_BasicType for SDR class, but keep the "hack" with isSparse from numenta/master (I'm afraid it was me who introduced it back then) Or we just do it right (tm)...and potentially break some things (let's keep that after #137 ) to see if the popular nupic .py repo is affected |
@dkeeney I think this was resolved by some recent PR (which?) Can we close this? |
Yes, it was resolved in #230. |
Objective:
Discuss removig
bool sparse
fromengine/Spec.hpp
;and possible replacement with defining NTA_BasicType_Sparse
to comply with OPF (py API)
discussed in #136
from @dkeeney 's comments: #136 (comment)
and followup #136 (comment)
The text was updated successfully, but these errors were encountered: