[Moin-user] Integrating external docs into moin

[Thomas Waldmann]

> > That's possible, but needs a rather dirty hack of Page/PageEditor
> > classes until the storage backend api is fully functional and merged
> > (moin 2.0).
> 
> Ah! So there'll be full support for custom backends in the not-too- 
> distant-future?

Well, it will likely take quite some months until we can merge that into
the main development repo. We first need to release 1.9 "the WSGI
release" (which is currently blocked by some auth / session brokenness),
then we can work on 2.0 "the storage release".

So it depends a bit on how many developers are helping with all this.

> That's exactly what I need.  Is this just planned, or  
> is it currently in development?

There was a SOC 2008 project (of Christopher Denter, mentored by
Johannes Berg) that successfully implemented most of the needed bits.
Pawel Pacana (mentored by Radomir Dopieralski) worked on a mercurial
storage backend using this API.

See there: http://hg.moinmo.in/moin/1.8-storage/

(1.8 in the repo name means it was based on the 1.8 main repo, this repo
was merged from christopher's and pawel's repo)

Here are some interesting pieces for you:

http://hg.moinmo.in/moin/1.8-storage/file/tip/MoinMoin/storage/backends/fs.py

My try at having server filesystem dirs / files accessible via the wiki
namespace (file browser within the wiki).

http://hg.moinmo.in/moin/1.8-storage/file/tip/MoinMoin/storage/backends/router.py

Middleware: Route requests to different backends depending on the item
name. So e.g. most pages are handled as usual by the wiki, pages FS/*
are directed to the fs backend.

That's both rather to be taken as an idea formulated in python than full
working code (see TODO there). If that stuff would fully work, it would
be a good base for what you need.

> Our wiki doesn't need to be ultra- 
> stable, so I'd be pretty happy running it on the bleeding edge code  
> from the repository if there's at least a partially-functioning  
> storage backend api existing.

As we currently still work on 1.9, code intended for 2.0 is very
bleeding edge. :)

Basic storage is working. router, fs, acl definitely need more work.

> Yeah, those are the sorts of problems I had in mind.  I have a feeling  
> that hacking Page/PageEditor would require less code changes, although  
> those changes would be much more fragile between releases.  I'll have  
> a play about and see how it works.

I guess it all depends on whether you want a q&d hack (soon) or a clean
solution (that requires more effort and time).

> Yeah, I'm planning to be there; it'll be good to catch up again.

I also plan to be there. Let's see. :)

> It's just a shame it's in Birmingham :)

I don't know Birmingham (nor other parts of England aside from
Heathrow), what's the problem?