Robert Sachunsky

> But it also installed `Keras 2.4.3`. And here is the result when running `sbb_textline_detector --help`: > > ``` > ImportError: Keras requires TensorFlow 2.2 or higher. Install TensorFlow via...

> > The only thing that troubles me with delegating to thin module Docker containers is that we more or less surrender version control. It's really difficult to do that...

Anyway, with #118 ff. the original issue (different TF requirements) has been solved. The topic has since moved on to how do we integrate/compose thin module Docker images (where available)...

> > And it is actually possible to do without over-engineering too hard. I.e. I use a `run` one-liner that just calls a container with the right UID and working...

Besides spinning up multiple CLI-only containers (somehow) sharing volumes, we could also integrate containers as true network services, but merely by installing a thin OpenSSH server layer on top of...

> ocrd-segment-repair has the optional operations "plausibilize" and "sanitize" – I have no idea what this exactly does :) I agree, these are not expressive enough, or even memorable (which...

> Documentation from https://ocr-d.de/en/workflows: > > * plausibilize = Remove redundant (almost equal or almost contained) regions, and merge overlapping regions > * sanitize = Shrink and/or expand a region...

I'm not sure #287 actually does what we want. This looks like we still have races: ``` sem --will-cite --fg --id ocrd_all_pip/usr/local/sub-venv/headless-tf1 python3 -m venv /usr/local/sub-venv/headless-tf1 sem --will-cite --fg --id...

The above was from three different variants of our CI. I am also seeing strange effects in native installations – if I enable `-j`, not all modules will be built....

Also, I think it would be useful to add an option for **not** moving any local files around at all, including ID changes. (In that case, no references need to...