Closed arademaker closed 3 years ago
Hmm, nothing seems particularly amiss at first glance. I do notice that in your command-line version you are parsing while in the Python code you are generating. Do you get the error at the command line when you generate?
Oh, my bad. I want to parse. I forgot to change the example from the documentation. But the generation should work right?
Yes, I think it should work. But I wonder if the source profile has an empty result
file, so there's nothing to generate from. If this is the case, there shouldn't be an exception, or at least a more informative one should be raised.
The source profile is the golden
at https://github.com/arademaker/sick-fftb. It was annotated with fftb.
Ok I tried it out. I added the following to process.py
so I could see the progress (the same as using -vv
when run from the command line):
import logging
logging.basicConfig(level=logging.INFO)
And I used the 2018 version of the ERG since that is what I had.
Parsing works for me (replaced ACEGenerator with ACEParser). In process.py
you don't constrain the number of results so it takes about a minute on my machine (compared to ~10s with -n1
), but there were no errors.
Generation also completes without issue, but it takes about 5 minutes.
In any case, the error you got looks like ACE crashed and then PyDelphin tried to send another item to the dead process. I'm not really sure that anything is wrong with PyDelphin in this case.
Closing as invalid; it looks like nothing was amiss here.
After finishing the annotation of https://github.com/arademaker/sick-fftb, I am using pydelphin to create a new profile with the best analysis for each sentence. The ideia is to compare the golden analysis with the best analysis from ACE. But...
the code is:
but from the command line, it works fine: