[question] How to best package an app that uses Tensorflow?

[Giovanni Campagna]

Hello all,


I am the maintainer of Almond (edu.stanford.Almond) upstream and on
Flathub.
The latest version of Almond would like to bundle Tensorflow, a python
library for machine learning / neural networks,
and I'm wondering what is the best way to package it.

The Tensorflow developers (Google) provide pre-built python wheels for
x86_64 and armv7 (for Raspbian). There is no source pip package.

Building Tensorflow from source is possible (I've done it) but it's
*very* slow - on a good machine it can take many hours. It also brings
in bazel, a Google-specific build system in Java. I've never tried
building bazel from source, and I don't know how easy that would be.

So I'm wondering: is it acceptable for Flathub to download the prebuilt
Tensorflow binaries for x86_64 and arm?
Should I specify them as plain sources or do I need to use "extra-
data"?
Can the 32-bit arm code be used by an aarch64 app if I include a 32-bit 
python?
Finally, has anyone else tried building Tensorflow on Flathub? If so,
how did you do it?


Thanks in advance for your answers and suggestions, and cheers,

Giovanni

PS: please keep me CCed on all your replies, I am not subscribed to
this list.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <https://lists.freedesktop.org/archives/flatpak/attachments/20190423/531de251/attachment.sig>