[Bf-docboard] blender wiki at Amsterdam Conference

[Dan McGrath]

Hey,

Just a quick little status report that seemed like a better fit as an
inline reply to this email instead of standalone. Enjoy!

On Tue, Oct 23, 2012 at 1:48 PM, Jim Tucker <jim.tucker at live.co.uk> wrote:
> Some discussion is going on in the wiki admin team, as important technical
> decisions
> have to be taken in order to keep the wiki up to date with less effort that
> it’s needed now. Hopefully within a
>  couple of weeks the situation will be clearer. The changes that need to be
> operated in the backend will most
> probably have no direct impact on current frontend/content wiki work of
> course.

As far as the backend goes, only some partial tests have been carried
out using a multi branch git repo to manage the MediaWiki installation
and deploy to the vhost directory using hooks. Things like file system
locations, permissions, naming conventions and hook code, for example,
need to have the details hammered out and finalized (Marco, got any
time this week?), which I think should be coordinated with the server
admin(s). At this point, even I am not exactly sure what the scope of
the changes I should be aiming for are (clarification here Ton?), I
was just handed the keys :D

That aside, the existing install still needs to be cleaned up and
organized. As a worst case, the upgrade could happen without moving
stuff to git, but I fear that leaving the current situation in place
could complicate things for future work on the system, especially if
the upgrade fails, or if the upgrade doesn't get committed to svn and
leaves the working directory even more in a state of drift than it
currently is. A db backup and quick copy of the htdocs for the vhost
would obviously be carried out before any changes, but still would be
nice to just Make It Right (tm).

Also, downtime needs to be announced to the ML and via the wiki's site
message (which I actually don't know how to control yet, Luca can help
here?), so that users know not to/can't edit stuff during the
maintenance window, just in case we did have to roll back and possibly
loose some edits that users may not notice. And in a worst case of a
failed upgrade/basic disaster recovery, how much down time is
tolerated by something like Google's indexing? Better to play it safe,
right? :) I don't think that we have anything in place currently to
switch over to any kind of emergency offline/maintenance mode, short
of hand editing the server vhost config and a quick new index.php
htdocs root designed to spit out 503's and retry-after's in case
something goes horribly wrong, as per:

  http://googlewebmastercentral.blogspot.ca/2011/01/how-to-deal-with-planned-site-downtime.html

I am hope that procedural changes like these would be put in place for
the longer term solution of managing the server. I guess until I get a
green light, and have a fully working and tested system in place that
the other admins can test and sign off on, I can't say for sure when
an upgrade will take place.

One of the things that I would like is a confirmation from the server
admin to clarify the backup situation to ensure the wiki is backed up
to a separate machine (or make it known if not so that preparations
can be made etc.), as stuff I would like to do is being developed on
the production system, and thus a little more risky (puppet someday
please!). I also wouldn't mind having a good chat with all of the
admins together at some point to go over some of the specifics, since
they would all have to be aware of the changes and (re)learn. So far I
have only heard back with general support of the svn to git
migration/cleanup idea. I am not sure if git is a welcomed choice, or
if perhaps mercurial would be better, or maybe even a cleaned up svn
again etc. But given that we are using MediaWiki which uses git now,
git seems like a good choice. There are lots of ways to do things
though.

Unfortunately, real life makes things a little difficult to coordinate
a bunch of volunteers with day jobs in different time zones to meet on
irc, and emails don't always get answered as fast or as thoroughly as
one would like compared to irc. But agreed that the server stuff
should really have no impact on the wiki content, other than perhaps
the odd change in extensions and features here and there.

> It is recognised that the wiki is a valuable and widely-used resource, and
> that a huge amount of effort has been, and still is,
> being put into the major changes needed following the 2.4 to 2.5 system
> change.  But it still needs a lot
> of work to get it to a fully-completed state, and until recently progress
> has been slow.  We hope that
> upcoming changes (new  Writer Guide, etc.) will increase the flow.  A call
> was made for volunteers
> to help with this effort. Unfortunately this received very few replies; but
> F. Siddi offered to put a
> call out on Blender Nation (which reaches some 2500 readers rather than the
> 200 Conference delegates) when
> Kesten is ready to start the next sprint.

Good luck with the call out, that sounds like a great idea!

> I think our efforts were under-represented at this Conference. I would like
> to suggest  that for the next Conference
> there should be a proper report  on the state of the Blender Manual, with a
> discussion of its merits, demerits and problems,
> and thanks to all the volunteer  writers/reviewers for their work.

Anyways, my email wasn't so much about the wiki content, but I figured
that I should at least let everyone know where things stand, and if
everyone approves, that I would be glad to work the admin(s) and the
community towards a nice tolerant little system for all to enjoy! :)

Please feel to share any experience, ideas or suggestions you have as
well, as I love hearing about such geeky topics! Later o/


Dan