Issue in running the script "twice"

[Diogo-Rossi]

Some times, there will be found another issue regarding the place where the run() function is called.

Imagine the following situation:

# Contents of `script.py`
from abaqus import *
from abaqusConstants import *
from driverUtils import *

# Do code stuff:
# * Read and generate data and make calculations 
# * Open, create, delete, move files and folders 
# * Change the CWD to launch Abaqus in separate folder
# * etc...

mdb.saveAs('model.cae')

In the above code, the run() function is called in the mdb.saveAs() method, which launches Abaqus and exit.

Because of that, if mdb.saveAs() is called at the end, the "code stuff" will run twice: by the python3 interpreter (with the command line python3 script.py) and by the abaqus python2 interpreter. And that could cause some issues in operations like writing files and moving folders.

One possible solution is:

Move the call of run() function to the modules abaqus and odbAcess.

When we import abaqus or import odbAcess, the function will launch Abaqus and exit, not running the "code stuff" in python3interpreter

Is that feasible?

[haiiliin]

I think when building the model, we would rarely write something, only when we are getting the output data.

If we do this, much of the code will not be executed in python 3 interpreter, what if we want to do some debugging in python 3 to see what is in the model, though it is different in the abaqus python 2 interpreter, we can still get some information about the model.

[Diogo-Rossi]

Oh, I got your point, that would really be useful.

But still: what if I want to do several launches and buildings with one single script? (I mean, create folders, moving to folders, etc)

Another thing is: the script will behave differently in python3 environment and in abaqus python2 interpreter, which may be inconsistent, as the point is to write the same script in an different environment, I think.

what if we want to do some debugging in python 3 to see what is in the model

That would be really useful! And I think this is a solution for that.

[haiiliin]

Oh, this should be useful.

[Diogo-Rossi]

With that, maybe we don't need to use exit_after argument in run() function. It could decide whether or not to exit (if in debuging or in normal run)

[haiiliin]

Right, excellent idea.

[Diogo-Rossi]

I was think about to move all the calls of the abqpy.abaqus.run() to only 2 places:

---

• End of file src/abaqus/___init__.py:

...
gettrace = getattr(sys, 'gettrace', None)
exit_after = False
if gettrace is None or not gettrace():
    exit_after = True

abqpy.abaqus.run(cae=True, exit_after=exit_after)

• End of file src/odbAccess.py:

...
gettrace = getattr(sys, 'gettrace', None)
exit_after = False
if gettrace is None or not gettrace():
    exit_after = True

abqpy.abaqus.run(cae=False, exit_after=exit_after)

---

In this way, both import abaqus and import odbAccess will run the abaqus command and exit in Python 3 interpreter
So, it will not run the rest of the script twice (because it will already run in Python 2).
Except for when in debug mode. In this case, Python 3 will run the file anyway.

What do you think?

[haiiliin]

We can define a environment variable to indicate whether to exit in abaqus and odbAccess module, the variable is True by default, in pytest, we set the variable to False.

This should be possible

[Diogo-Rossi]

So, how to set the variable to False only in pytest?

[haiiliin]

Possibly in the pytest_configure function, use os.environ to set environment variable

[haiiliin]

abqpy/src/abqpy/abaqus.py

Lines 33 to 43 in 84f7eec

	# check if in debug mode and run
	debug = os.environ.get("ABQPY_DEBUG", "False").lower() == "true"
	print('Debug mode is on because ABQPY_DEBUG is set to true')
	gettrace = getattr(sys, 'gettrace', None)
	if not debug and gettrace is not None and gettrace():
	debug = True
	print("Debug mode is on because of debugger")
	# if not in debug mode, exit the program
	if not debug:
	sys.exit(0)

abqpy/tests/conftest.py

Lines 9 to 21 in 84f7eec

	def pytest_configure(config: pytest.Config) -> None:
	"""Allow plugins and conftest files to perform initial configuration.
	This hook is called for every plugin and initial conftest file after command line options have been parsed.
	After that, the hook is called for other conftest files as they are imported.
	Args:
	config (pytest.Config): The pytest config object.
	"""
	os.environ['ABQPY_DEBUG'] = 'true'
	os.chdir(os.path.dirname(os.path.abspath(__file__)))
	sys.path.insert(0, os.path.abspath('../src'))

This should solve the problem

[haiiliin]

abqpy/src/abqpy/abaqus.py

Lines 29 to 41 in e7ed5b2

	# check if in debug mode and run
	debug = os.environ.get("ABQPY_DEBUG", "False").lower() == "true"
	gettrace = getattr(sys, 'gettrace', None)
	if not debug and gettrace is not None and gettrace():
	debug = True
	# If in debug mode do not run the abaqus command at all
	if not debug:
	if cae:
	os.system(f"{abaqus} cae noGUI={filePath} -- {args}")
	else:
	os.system(f"{abaqus} python {filePath} {args}")
	sys.exit(0)

And I think we should not run the Abaqus command at all if it is in debug mode.

[haiiliin]

Implemented in #1501.