[Qgis-developer] Wiki page/area for 'Building QGIS from scratch'

[John Hawkinson]

Replies to multiple messages inline, elow.

Tim Sutton <tim at kartoza.com> wrote on Tue, 21 Mar 2017 at 21:43:07 +0200:

> John have you considered just running QGIS in docker - its probably
> a much easier approach than the method you have embarked upon...

I think your question was directed to Mark Johnson, not to me? But since
you asked, I found the rational raised in this thread to be...stunningly
unconvincing!:

Nathan Woodrow <madmanwoo at gmail.com> wrote Sat, 18 Mar 2017 at 22:40:21 +1000:

> The wiki information got outdated very fast,

The reason documentation gets outdated is typically because of a lack
of maintainers (possibly related to lack of visibility), and also an
overly high bar to editing it. A wiki is generally the lowest bar
possible. If the problem is information getting outdated quickly,
moving away from a wiki seems a step in the wrong direction, not the
right one.

> and was also not translatable easily.

I'm not very familiar with translation issues, but my understanding is
they aren't usually strictly format-dependent. (But I should
acknowledge my deep biases as an English speaker.)

> A decision was made a while ago for official documentation should be
> done int the same workflow as the website in order to streamline
> that process

I'm not sure why a streamlined website process ought to have
any bearing on the quality or accessibility of documentation. This
seems a second-class issue.

> and make sure everyone is using the same tools.

I'm also not sure I understand this or why it is important for
documentation; or more important than good documentatoin.


Richard Duivenvoorde <rdmailings at duif.net> Sat, 18 Mar 2017 at 15:55:34 +0100:

> And can I add the constant hard fight against fake user accounts adding
> wiki spam? (see osgeo wiki).

Yes, this I understand; and this is the constant tension between
lowering the barrier of entry far enough to make it easy for casual
contributors to crowdsource information, and lowering it too far to
make it too easy for spammers. Except it seems like the osgeo wiki
has finally evolved a fairly effective solution (modulo the fact that
it's awful hard to get a mantra and credentials!).

> The workflow we have now is translatable, can be build to pdf's, is
> versioned and is continuously tested/rebuild.

Continuous testing is super-helpful for code. It's nice for
documentation, but not really in the same way. I'd much rather have
good documentation than continuously-tested documentation.

Maybe that's an artificial tradeoff and we can have both.
But I'm not convinced (not that I need to be; again, I'm new here and
so far mostly send long emails.)

--jhawk at mit.edu
  John Hawkinson