ArcGIS Pro v2.4

2203
5
Jump to solution
03-19-2019 07:35 PM
DanPatterson_Retired
MVP Emeritus

I know that 2.4 is coming, but I needed new functionality and some upgrades in numpy, scipy.  I upgraded and have found no issues so far, but I was wondering if the alpha/beta team is using the new releases with it and whether anything on the list raises a flag that I should watch out for.

Here is the conda update and the module changes

(arcgispro-py3) C:\arc_pro\bin\Python\envs\arcgispro-py3>conda update numpy --no-pin
Fetching package metadata .............
Solving package specifications: .

Package plan for installation in environment C:\arc_pro\bin\Python\envs\arcgispro-py3:

The following packages will be UPDATED:

    icc_rt:     2018.0.2-arcgispro_0  esri [arcgispro] --> 2019.0.0-h0cc432a_1
    mkl:        2018.0.3-arcgispro_0  esri [arcgispro] --> 2019.1-144
    mkl_fft:    1.0.6-py36hdbbee80_0       --> 1.0.10-py36h14836fe_0
    mkl_random: 1.0.1-py36h77b88f5_1       --> 1.0.2-py36h343c172_0
    numexpr:    2.6.8-py36h9ef55f4_0       --> 2.6.9-py36hdce8814_0
    numpy:      1.15.4-py36ha559c80_0      --> 1.16.2-py36h19fb1c0_0
    numpy-base: 1.15.4-py36h8128ebf_0      --> 1.16.2-py36hc3f5095_0
    scipy:      1.1.0-py36h4f6bf74_1       --> 1.2.1-py36h29ff71c_0‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

UPDATE

Upgrade documentation, the 1.16 series is the last to support python 2.7

numpy/doc/release at master · numpy/numpy · GitHub   General

numpy/1.16.0-notes.rst at master · numpy/numpy · GitHub  specific to 1.16. further 'dot' modifiers are bug fixes largely

Scipy dumps 2.7 after this release as well.  Plus lots of improvements

scipy/1.2.0-notes.rst at master · scipy/scipy · GitHub 

1 Solution

Accepted Solutions
DanPatterson_Retired
MVP Emeritus

Looks like a 'not' on my question

From here

https://anaconda.org/Esri/repo  and specifically here https://anaconda.org/esri/ArcGISpro 

Did a dry-run and got one new package (hdf5) and a bunch of downgrades.

This may not be the final version to be released prior to beta_2.4, but it will give you an idea I suppose

So make sure you at least keep a list of the packages you updated prior to doing an install of 2.4 when it is released for beta.  In my case none of the packages I needed upgraded were done (numpy 1.16.2 and spyder 2.3.3 or 2.3.4 amongst a few).  You will need to 'freshen up' your clone or the default env depending on how you install ArcGIS pro

Kory Kramer‌, I suspect everyone is too busy with 2.4 pre-release to add any comments

(arcgispro-py3) C:\arc_pro\bin\Python\envs\arcgispro-py3>conda install -c esri arcgispro --no-pin --dry-run
Fetching package metadata .............
Solving package specifications: .

Package plan for installation in environment C:\arc_pro\bin\Python\envs\arcgispro-py3:

The following NEW packages will be INSTALLED:

    hdf5:       1.10.2-hac2f561_1

The following packages will be UPDATED:

    arcgispro:  2.3-h35c310c_0         esri --> 2.4-0                 esri

The following packages will be SUPERSEDED by a higher-priority channel:

    icc_rt:     2019.0.0-h0cc432a_1         --> 2018.0.2-arcgispro_0  esri [arcgispro]
    mkl:        2019.1-144                  --> 2018.0.3-arcgispro_0  esri [arcgispro]

The following packages will be DOWNGRADED:

    h5py:       2.8.0-py36_arcgispro_4 esri [arcgispro] --> 2.8.0-py36h3bdd7fb_2
    mkl_fft:    1.0.10-py36h14836fe_0       --> 1.0.6-py36hdbbee80_0
    mkl_random: 1.0.2-py36h343c172_0        --> 1.0.1-py36h77b88f5_1
    netcdf4:    1.4.1-py36_arcgispro_3 esri [arcgispro] --> 1.3.1-py36_1          esri [arcgispro]
    numexpr:    2.6.9-py36hdce8814_0        --> 2.6.8-py36h9ef55f4_0
    numpy:      1.16.2-py36h19fb1c0_0       --> 1.14.6-py36hc27ee41_4
    numpy-base: 1.16.2-py36hc3f5095_0       --> 1.14.6-py36h8128ebf_4
    scipy:      1.2.1-py36h29ff71c_0        --> 1.1.0-py36hc28095f_0

View solution in original post

0 Kudos
5 Replies
DanPatterson_Retired
MVP Emeritus

Kory Kramer‌ can you ask the development team?

https://anaconda.org/intel/icc_rt is the fortran compiler used for something specific

and are the math libraries

Intel® Math Kernel Library (Intel® MKL) | Intel® Software  and mkl-fft replacements for something old or what new functionality is offered.

don't see any references in the help regarding these

0 Kudos
KoryKramer
Esri Community Moderator

Shaun Walbridge‌ would you be able to help here?

0 Kudos
DanPatterson_Retired
MVP Emeritus

10 days of testing and I have had no issues.  Lots of viewers however!  Taking a wild guess that nothing is going to change until the anaconda distribution with python 3.7 upgrades their numpy and scipy versions as well.

For those that can't wait, or want to tempt fate, just provide version checks if you share your code with others.

I just now provide version checks until the python, numpy and scipy versions in Pro catch up.

As for people working with structured/recarrays in numpy and arcpy, there are some useful additions to numpy.

A version check for people working with versions < 1.16 can be done simply for some functions.

 I just add any needed checks in subsequent lines for other numpy or scipy changes or additions in subsequent lines

v =  np.version.version.split('.')[1]  # version check
if int(v) >= 16:
    from numpy.lib.recfunctions import structured_to_unstructured as stu

Similarly for python

import sys
sys.version

'3.6.8 |Anaconda, Inc.| (default, Dec 30 2018, 18:50:55) [MSC v.1915 64 bit (AMD64)]'
DanPatterson_Retired
MVP Emeritus

Bump... Lots of esri staff viewing the thread, but not confirmations as of yet for anything.

Also to add to the list, Spyder 3.3.4 is to be out through the anaconda distribution in a couple of days.  

Is the Pro 2.4 package set frozen during the pre-beta stage or will it be updated before the final release around the User Conference?

0 Kudos
DanPatterson_Retired
MVP Emeritus

Looks like a 'not' on my question

From here

https://anaconda.org/Esri/repo  and specifically here https://anaconda.org/esri/ArcGISpro 

Did a dry-run and got one new package (hdf5) and a bunch of downgrades.

This may not be the final version to be released prior to beta_2.4, but it will give you an idea I suppose

So make sure you at least keep a list of the packages you updated prior to doing an install of 2.4 when it is released for beta.  In my case none of the packages I needed upgraded were done (numpy 1.16.2 and spyder 2.3.3 or 2.3.4 amongst a few).  You will need to 'freshen up' your clone or the default env depending on how you install ArcGIS pro

Kory Kramer‌, I suspect everyone is too busy with 2.4 pre-release to add any comments

(arcgispro-py3) C:\arc_pro\bin\Python\envs\arcgispro-py3>conda install -c esri arcgispro --no-pin --dry-run
Fetching package metadata .............
Solving package specifications: .

Package plan for installation in environment C:\arc_pro\bin\Python\envs\arcgispro-py3:

The following NEW packages will be INSTALLED:

    hdf5:       1.10.2-hac2f561_1

The following packages will be UPDATED:

    arcgispro:  2.3-h35c310c_0         esri --> 2.4-0                 esri

The following packages will be SUPERSEDED by a higher-priority channel:

    icc_rt:     2019.0.0-h0cc432a_1         --> 2018.0.2-arcgispro_0  esri [arcgispro]
    mkl:        2019.1-144                  --> 2018.0.3-arcgispro_0  esri [arcgispro]

The following packages will be DOWNGRADED:

    h5py:       2.8.0-py36_arcgispro_4 esri [arcgispro] --> 2.8.0-py36h3bdd7fb_2
    mkl_fft:    1.0.10-py36h14836fe_0       --> 1.0.6-py36hdbbee80_0
    mkl_random: 1.0.2-py36h343c172_0        --> 1.0.1-py36h77b88f5_1
    netcdf4:    1.4.1-py36_arcgispro_3 esri [arcgispro] --> 1.3.1-py36_1          esri [arcgispro]
    numexpr:    2.6.9-py36hdce8814_0        --> 2.6.8-py36h9ef55f4_0
    numpy:      1.16.2-py36h19fb1c0_0       --> 1.14.6-py36hc27ee41_4
    numpy-base: 1.16.2-py36hc3f5095_0       --> 1.14.6-py36h8128ebf_4
    scipy:      1.2.1-py36h29ff71c_0        --> 1.1.0-py36hc28095f_0
0 Kudos