[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Missing dependancies for streamlit





-- 
Michael R. Crusoe

On Tue, Apr 7, 2020, 03:27 timsn_thtree_net <timsn@thtree.net> wrote:
Hi Andreas,

Not certain what distro/config streamlit was made in, but the Makefile
assumes that pip = pip3, and it looks like they really want python 3.8,
but things seem to move ahead with the 3.7.3 version installed.

Going past that by tweaking the makefile there are some things to
install and some things missing in Debian, as those are outside of
Debian Med, not certain who coordinates those other python packages.

Missing stuff:

Virtualenv location:
Installing dependencies from Pipfile…
An error occurred while installing cachetools>=4.0! Will try again.
   🐍   ▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉ 47/47 — 00:01:01
An error occurred while installing tensorflow>=2.0.0; python_version <
'3.8'! Will try again.
An error occurred while installing mypy==0.761! Will try again.


https://packages.debian.org/search?keywords=python3-cachetools

v4.0 is only in bullseye testing

You should setup a Debian unstable package building environment using pbuilder or should. This is required and will also solve some of these package version issues.


I would patch the setup.py to accept the newer mypy. It is unlikely they need it for running or building, so you may be able to patch it out completely (and maybe ask the upstream authors to make it optional)



tensorflow 2 apparently requires a pip3 version of > 19 according to
https://www.tensorflow.org/install, but current buster only has 18.1

Since you will have to manually ensure that all the right dependencies are available it shouldn't be a problem to force the use of a different pip version.

However, Debian packages are introduced to the "unstable" distribution first, which has the newer pip version as you noticed, so you should be fine.



https://packages.debian.org/bullseye/python3-pip says bullseye testing
has 20.0.2-3


I can try to install those versions of cachetools and mypy, then see
what how far I get with tensorflow2 (having previous played a bit with
tensorflow, I know it can be a bear for dependancies).

Great, thank you!




TTUL,

TJ


Reply to: