Robert Sachunsky

> I don't know what's the correct way to deal with this, but it seems to suffice checking that a `.__file__` exists at the module level – for namespaces this...

CI failure is only on macos, and [very strange](https://app.circleci.com/pipelines/github/OCR-D/core/1389/workflows/9d0c400e-630c-4761-b4a6-bc8e0702245b/jobs/4616/steps): all tests seem to pass, but the overall status is still failure. I don't have permissions to rerun on CircleCI.

Also, I believe it is not correct that `generate_range` greedily selects the first numerical substring. Page identifiers could be made up of several numbers...

> I'm missing the word `region` in the parameters (regions→lines) Why? You'd only need that for region segmentation (page→regions). The two paragraphs above the one you quoted clearly explain that....

> `highest level` = something like _region_ and `lowest level` = something like _glyph_? yes > and `to segment` = _to be segmented_ or _to be the result of segmentation_?...

Valuable input: https://hackmd.io/g-8QuOAnSQeqnb8PRNjFbQ

I am quite surprised to (re-)discover [the existing OCR-D developer guide](https://github.com/OCR-D/ocrd-website/blob/master/site/en/developer-guide.md). This seems to be buried in the repo (without a link on the website) and slightly out of date...

Also: templates for processors of various task groups (preprocessing, segmentation, recognition, postprocessing). Or at least runnable API examples.

[is what I'm sayin all along](https://github.com/OCR-D/ocrd-website/issues/98) :grin:

@kba @lena-hinrichsen this should be a high documentation priority. It is highly confusing to users that they not only need to type long docker run calls every time (when what...