Split jekyll import into separate repo / binary

[Geobert]

Yeah, it was my idea and my first contribution to the project, but the more I look at it the more I think it was not a great idea: 

• No other static generator is doing it
• My knowledge of Jekyll is not good at all
• The current state of the converter is bad: the produced files are not usable by cobalt, leading users into a bad experience
• It make cobalt bigger in term of codesize for a feature that is lightyear to maturity

Hence this ticket: shall we get rid of the feature before too many people notice it? Or deactivate it? Or at least print a giga banner saying this is pure experimental and is not working yet?

[berkus]

Hey, I am in favor of dumping the jekyll importer and focusing on improving the current cobalt model.

Only thing I want from the jekyll implementation is the frontmatter delimiter syntax as indicated in #431.

[Geobert]

I've seen this yeah, I agree with your proposal but the change is not that easy to do: the code has to be modified (following edpage proposition), new tests must be created and all. So contributions are welcome :)

[berkus]

Sure, if there's agreement from maintainers that this should be done I can invest some time in porting it from jekyll-model.

[Geobert]

I maybe wrong, but I think jekyll-model module was a refactor of my initial contribution to convert jekyll's site into cobalt's. So to support the double separator, we need to modify https://github.com/cobalt-org/cobalt.rs/blob/master/src/cobalt_model/document.rs#L46

[epage]

@Geobert
I think its a good thing we have improters. I eventually plan to write an evernote import tool.

However, I could see value in pulling the import tools out into separate [bin]s / repos. That will better allow independent evolution (like bumping major) and better tracking of support / behavior.

In general, we've taken the approach of cobalt being a completely self-contained binary. I think its an acceptable exception for import tools. They are a one-off use.

@berkus

Sure, if there's agreement from maintainers that this should be done

Sorry I wasn't clear. My comment in #431 and marking it as an enhancement were acceptance of it. I'll gladly accept a PR that makes that change.

[Geobert]

Good idea this split!