Re: Software overhaul? - Frontend experience

Tobie, I realized that I had some thoughts on a thread with *you *back in
the day about this. The subject line is "Plan C".

For everyone else's benefit, here are my thoughts from back then:

************

I didn't know much about Gollum before (I've never used GitHub's wikis),
but after a bit of research and playing around with Gollum, here are my
(very, very, disorganized) thoughts in the interest of spurring discussion:

I *really **like*:

   - Static files are much, much easier to find hosting for
   - Sends a very strong signal that this is a community-owned endeavor
   - Being able to clone the wiki for offline consumption is nothing to
   sneeze at.  I can't tell you the number of times I've been hacking
   something together on a plane and gotten frustrated from having to go off
   of my memory of which CSS values I can give for a property or something
   - A lot of flexibility to add our own features over time, or for the
   community to

I'm *not such a huge fan of*:

   - We want people to actively participate in the wiki. Requiring people
   to log in is already a huge barrier; compared to that, requiring folks to
   know git, clone the repo, and issue a pull request seems like an *
   impossibly* high barrier
   - No transclusion ability (no templates)


I don't know how exactly we'd run something like this in practice.  Here's
what I'm imagining based on my extremely limited (and likely incorrect)
understanding of Gollum:


   - *I'm assuming the simple web-based Gollum editor doesn't scale very
   well, is hard to permission, etc.*
   - We have a small server running the web-based gollum editor running
   somewhere off the beaten path, like edit.webplatform.org
   - Most traffic goes to the static docs.webplatform.org, which is hosted
   on something like Google App Engine
   - On each wiki page on the static site we have a link to the equivalent
   page on edit.webplatform.org. Folks can edit any page they want using
   the web-based editor, and preview their changes there.
   - Every hour or so a cron job deploys the most recent version of the
   gollum wiki to the main, static serving system, based on what's in
   edit.webplatform.org
   - If someone wants to do something more involved they can clone the repo
   and issue a pull request, which one of the admins approves.



On Wed, Jun 12, 2013 at 12:27 AM, Chris Mills <cmills@opera.com> wrote:

>
>
> On 7 Jun 2013, at 15:08, Jonathan Garbee <jonathan.garbee@gmail.com>
> wrote:
>
> > The current state of WPD in terms of software is very poor. A lot of
> issues have been fixed over the last few months, but overall the system as
> a whole isn't up to what most of us seem to think the standards should be.
> A few of us have been experimenting with different platforms and seeing if
> they would meet what we think WPD would be. However the progress has been
> seemingly slow and without real community discussion of what the standards
> should be that are being thought about. That is changing.
> >
> > We need to discuss what the contributor experience should be like and
> formalize a set of goals for how everything will work and interact. This
> way as we are analyzing possible solutions, we can find the one that best
> suites those end goals. After a few months of being live we have spotted
> many pain-points for contribution, so taking that knowledge we need to
> figure out the most fluid way for contributors to work because what we have
> now simply doesn't cut it.
> >
> > What do you all think that the contributor experience should be like? If
> you can point to example pages online that show the workflow you're
> thinking of. Remember that this conversation is not about backends at all,
> just about how we think contributors should interact from the front-end.
>
> I think how it currently is, is now not too bad, although I worry that the
> sessions issue was fixed too late, and has put a lot of people off. It
> still occasionally rears its ugly head for me, for example last week it
> lost we two hours work, which annoyed me immensely.
>
> Other points:
>
> * While the site is now functional, it still feels very clunky, with long
> loading/saving times. I'm not sure how much we can do about this, whether
> it is just unavoidable to some extent for a site of this magnitude.
> * It needs to accept HTML as the input language. I'm starting to find the
> Wiki markup to be a barrier, even though I used to like it.
> * We need to make sure the templates are all present, and other such holes
> are plugged.
> * I'm still a bit worried about the beginner's material, and beginners
> finding their way to material that will help them, but I can accept that we
> need to do the CSS Props first. Then we can concentrate on beginner's stuff.
>
>
>