Initial Implementation of a Data class #177
Conversation
|
Yes, as far as I understand it, |
…to feature/1.0 Conflicts: exa/util/constants.py
tjduigna
changed the title
|
I updated #167 to reflect the development here. |
exa/core/data.py
Outdated
| else: # assume this is a function | ||
| source = getattr(mod, obj) | ||
| except Exception as e: | ||
| print("attempt to import source failed") |
There was a problem hiding this comment.
Should we log or print or warn the traceback/exception?
There was a problem hiding this comment.
Yeah just printing is lazy. Should we go with self.log.error(e) to get a one-liner of the exception or self.log.exception which will give us the stack trace as well?
| raise TraitError("source must be callable") | ||
| return prop['value'] | ||
|
|
||
| @observe('source') |
There was a problem hiding this comment.
I really like this use of these event triggers.
| *mod, obj = source.split('.') | ||
| mod = importlib.import_module('.'.join(mod)) | ||
| # TODO : do we also support providing init | ||
| # args and kwargs in case source is class? |
There was a problem hiding this comment.
In the tests I see source_args and kwargs. Were those intended to be passed to the constructor of the class? Is functools.partial something we could use here to ship args for callables (taking a bit of functional programming possibly)?
There was a problem hiding this comment.
Ahh yes this is something I mentioned in #167 perhaps we should move the design discussion to the Issue. I will post a response there.
| @@ -0,0 +1,14 @@ | |||
| FROM tensorflow/tensoflow:2.0.0-py3 | |||
There was a problem hiding this comment.
This file is part of #176 right?
Closes #167