On Sat, 2007-07-14 at 13:49 -0400, Paul W. Frields wrote:
> ## Clarification -- staging in this case means, changes
to a
> document are not shown live until the workflow on the new staged
> version is complete
So it's kind of like a private branch that gets merged back to (or
replaces) the live version, if I get your drift.
That is a fairly good analogy.
> ## KW - Yes, I think so. As long as an admin can force
an
> update to content via some reasonable method for the rare
> occasions (defacing, security risk, data risk, etc.), I think it
> is reasonable to keep a document published until it has passed
> through a QA and translation stage.
This gives us the flexibility of "right away publishing" from wikiland
but with better possibilities for l10n, PDF, etc.? OK, I'm down with
that.
Well, not really the flexibility for normal cases. That is, a document
can be visible at various stages in its workflow, but whether fresh
changes are shown lives depends on which stage it is and what it has to
go through to be live.
The idea is, the closer you get to a final, formal publishing (to
docs.fp.o/document-name/), the more restrictive each succeeding workflow
step is. Early draft flows easily, but once you start to e.g. get
things being translated, it is more difficult (in terms of process, not
tools) to make an edit.
> 3) at draft, i assume all members can edit
>
> ## KW - Yes. We want all members of Fedora Docs to be able to
> edit in this workflow.
+1
> 4) at stable, i assume only key members can edit... and this
> is where we go to translations
>
> ## KW - This makes sense. Translators find mistakes or have
> questions/clarifications for the original; an editor or original
> owner needs to be able to fix those and push the content back
> for translation (generate a new POT, alert translators with open
> PO files, whatever.) How much of this can be done through
> Plone? Through CVS?
The more we can do through Plone, the easier it will be for non-techies
to help with documentation. That's a major blocker for our subproject
right now -- I suspect there are people out there who can correct
grammar and write good standard English, but who are frightened of doing
CVS because they think they might irreparably break something. (A silly
or at least misplaced fear, but you have to respect the fact that they
care.) So if they can do this with a pleasant, or at least
comprehensible, web interface, that's a win in my book.
+1
This particular question around translators might be a bit different.
By the time a translator has found an error, the content is in the
'stable' workflow state, and resolution needs to be handled by a more
experienced team member (editor, writer) with sufficient permissions. A
decision has to be made (via bugzilla, mailing list) and a new POT
pushed (if needed), then some social work has to be done (alert
translators via mailing list, in addition to any workflow alerts they
receive.) In some cases, the error report was on a mailing list, where
the solution gets worked out, and the thread needs to be closed.
So, there is going to be complexity and 'harder' here because of the
number of people affected. But I do agree that the tooling shouldn't be
difficult to match -- i.e., don't make an editor drop to the CLI to
handle this. BUT -- the editor should be able to do this from the CLI
(cvs + Jon's daemon). And there should be sufficient
checks-and-balances to ensure other contributors are not walked on by
the changes. :)
Yes, we certainly don't want Plone to become unresponsive every
time the
CVS server is busy. :-) I could see being able to email people results,
such as a PDF document, might be useful or desirable some day.
This is essentially the autobuild server I was asking about. It can be
hacked on to do all sorts of stuff we might find useful. For example,
I'd like RelEng to be able to trigger a build and staging of all they
need for packages that have content originating in /cvs/docs (such as
the release notes, but could be parts of /usr/share/doc/, etc.).
> 6) does *every* edit go back into CVS.. or do we want to
> have inner plone staging with another state that is "commit"
> ... so we have to exclusively commit back to CVS
>
> ## KW - once a document is in our custom workflow, it should
> have all edits go to CVS. For example, a writer or
> editor/reviewer needs to be able to open the document as a
> direct file in their $editor. All edits have to be synced to
> the same control management, once it is in formal document
> space.
>
> Not formal doc space == Wiki, personal Plone space,
>
fedorapeople.org
>
> Formal doc space == XML builds into HTML hosted via a CMS
> (Plone)
Right. Part of the reason we want this is for good interaction with
Dimitris' DL work, where fixes can be detected visually by L10N'ers...
or by people who like detecting automagically against
fedora-docs-commits.
Detected one time and searched for in multiple instances?
http://translate.fedoraproject.org/search
This does mean we have to get the Kupu-fu right,
or at least to a point where people like Karsten, me, Dimitris, John, et
al., can add styles to Kupu as needed to support all elements we use in
DocBook XML.
This is interesting to ponder. We'll have to first decide how we are
going from XHTML => XML, then test what that does with the various
styles. I reckon we can do a fair bit with CSS classes. Might need a
bit of custom work in between (XHTML => XML w/ classes => XML) to
interpret the classes into the specific XML elements. Or, you know,
maybe it will have been done by someone already?
- Karsten
--
Karsten Wade, 108 Editor ^ Fedora Documentation Project
Sr. Developer Relations Mgr. |
fedoraproject.org/wiki/DocsProject
quaid.108.redhat.com | gpg key: AD0E0C41
////////////////////////////////// \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\