Robert Sachunsky
Robert Sachunsky
@kba I don't know TBH. It would help if someone else thought it through. (Perhaps we are missing something here.)
> > What is still left to do beyond #688 and how can I help? > I don't know TBH. It would help if someone else thought it through. (Perhaps...
> Can't we ~~not~~ try to bring the coordinate logic from tesserocr into OCR-D/core first, while working towards PAGE-level methods? But the exception was merely an experiment. We don't want...
> Thinking about this from the PAGE model is a different picture, though. Functions like `add_TextRegion` would have to throw some `CoordsNotInParentException` which needs to be handled differently in the...
For the grace period, I suggest building Python 3.7 images in addition to the current Python 3.6 based images for exhaustive testing in parallel to normal updates. Changes: - Makefile:...
Related: https://github.com/OCR-D/spec/issues/119
> Since we require `README.md`, this could be implemented right away. But modules usually don't bundle it into their Python distro, do they? If they packaged it under [PEP566](https://packaging.python.org/specifications/core-metadata/) _Description_...
Cf. #623 for a first partial attempt.
> I did work on an `OcrdMetsFilter` class for more fine-grained control for cloning workspaces but it was supposed to be used wherever METS is to be queried, cf. #582....
Just to add another use-case to the scenario: even a simple `OcrdMets.add_file` can become inefficient on large workspaces. (Becoming as slow as 1 op/sec.) The reason is that it looks...