Network Task RunScript NumPy

[SamT]

Hey Everyone,

Hoping somebody may have a solution to an error that I am facing.

I want to use numpy in an external script (test_numpy.py) as a network task called from another script (network_process.py)

The script runs successfully when launched as a task through the GUI but blocks indefinitely when called from python.

Stopping the server causes the script to run successfully. Removing "import numpy" also succeeds.

I have attached both scripts.

Metashape version is 1.6.2

[SamT]

Thinking that this it GIL related. See https://lists.gt.net/python/dev/1406603

Unfortunately this is stopping me from using any libraries that depend on numpy like Tensorflow.

[Alexey Pasumansky]

Hello Sam,

Can you please check if the following modification in your script solves the problem.

Instead of line 12: n_t.frames.append((chunk.key, 0)) put the following:

Code: [Select]

if t.target == Metashape.Tasks.ChunkTarget:
        n_t.chunks.append(chunk.key)
if t.target == Metashape.Tasks.FrameTarget:
        n_t.frames.append((chunk.key, 0))

[SamT]

Hi Alexey,

Issue persists unfortunately.

This runs fine:

Code: [Select]

# import numpy

print('RunScript complete')

This blocks indefinitely:

Code: [Select]

import numpy

print('RunScript complete')


[SamT]

I should also mention that this runs:

Code: [Select]

import requests

print('RunScript complete')

However it throws this Exception.

Code: [Select]

2020-04-29 11:23:43 Exception ignored in: <module 'threading' from 'C:\\Program Files\\Agisoft\\Metashape Pro\\python\\lib\\threading.py'>
2020-04-29 11:23:43 Traceback (most recent call last):
2020-04-29 11:23:43   File "C:\Program Files\Agisoft\Metashape Pro\python\lib\threading.py", line 1283, in _shutdown
2020-04-29 11:23:43     assert tlock.locked()
2020-04-29 11:23:43 AssertionError:


So I know that I can use libs I have installed with pip

[SamT]

Hi Alexey,

Did you manage to replicate this behavior or do you think it is just my system?

I have attached a minimum example.