Trc file writer add version param

[pkess]

While i tried to update and finish the changes in PR 1702 i saw some basic change that are quite usefull but are not the trace file version 1.1 changes itself. I collected them here now.

I also added the docstring as it was requested by @zariiii9003 .

I will create another PR for the addition of the TRC version 1.1 itself.

[pkess]

I tried to but i was not able to fix the sphix build. It looks like sphinx cannot resove the TRCFileVersion to can.TRCFileVersion.

Can anyone help me?

[zariiii9003]

I tried to but i was not able to fix the sphix build. It looks like sphinx cannot resove the TRCFileVersion to can.TRCFileVersion.

Can anyone help me?

Sphinx fails to create a link to TRCFileVersion, because it does not exist in the documentation yet. You could add

.. autoclass:: can.TRCFileVersion
    :show-inheritance:
    :members:
    :undoc-members:

somewhere in file_io.rst.

[pkess]

Thank you for that hint. I will try that later

[zariiii9003]

instance attributes should be set in __init__. You could have version and format string as return values of a static method or just remove this method and put the try/except block into __init__

[pkess]

ok. The branch where i picked this from handled it with a tuple returned. I think this does not improve readability.

As this method is invoked from the init function it is always ensured that the attributes are initialized. Can you tell my why it should be added explicit in init?

[zariiii9003]

It's just a good practice, and it helps with code completion since IDEs usually look into __init__ for instance attributes.

I wonder why pylint is silent, there's a rule for this: https://pylint.readthedocs.io/en/latest/user_guide/messages/warning/attribute-defined-outside-init.html 🤔

[pkess]

I googled a bit for this and it seems like the only requirement is that all attributes are already initialized after the init function was executed. It is not enforced that the assignment is inside of the init function. So for me it looks like the approach here is correct.

I prefer to try to make a function implementation fit on one or two screen page. With this implementation we are quite prepared for more additions and changes not to blow up the function.

If we return a tuple we would achieve the same that is correct, but i think it is better readable like this.

So please take another look and tell me your thoughts.