On Thu, May 28, 2015 at 6:34 PM, Roman Shaposhnik <[email protected]> wrote: > On Thu, May 28, 2015 at 8:53 AM, David Yozie <[email protected]> wrote: >> If we want to continue using the same publishing system we should >> probably manage the end-user docs source in a separate branch of >> whatever repo it lives in (that's assuming we can't have a dedicated >> [...] > > The current publishing system is OK for what it does, but it reduces > our friendliness towards downstream consumption. Consider, for example, > if we want Geode to be packaged by Linux distros (or Bigtop). That > would require us to produce a -doc RPM/DEB to allow for a standalone > doc packaging. If we have extra requirements like Ruby, etc. that makes > it slight more complex. > > David, what are the chances of moving docs to a model where static > html pages can be generated?
Roman: The ruby-based "bookbinder" publishing system that we describe in the readme is really just one of many possible options for producing final documentation from the source files. The documentation is sourced in DITA format, and the DITA open toolkit (DITA-OT) performs the bulk of the conversion. Using only Java tools and the DITA-OT, you can output to a variety of different formats (see http://www.dita-ot.org/2.0/index.html for a list). That would be the route for generating plain HTML files, eclipse/HTML help, or whatever other format a distro might want to use. Having said that, I still think we should continue maintaining the source DITA files in a way that let's us easily use bookbinder and the CI system that builds professional output for the latest geode docs (http://geode-docs.cfapps.io/). -David > > Thanks, > Roman.
