Archive
Externalized dependencies
Comment on incremental builds
Principles:
- Retain as much of the formatted output as possible from build to build.
- Maintain an individual build stream for each topic from souce to output. Where merge is necessary as during book processing, defer merge as late as possible.
- Save every stage of the build stream as an intermediate file so that the build can rebuild as late in the build stream as possible.
- Separate the aspects of building a topic that are specific to each map. Also separate the aspect of building a topic that is independent of any map.
- Defer conditional processing as late as possible to minimize rework prior to filtering.
- Use timestamps and deltas to determine when to rebuild.
While a full rebuild with these principles will take longer, an incremental rebuild should take much less time. For instance, moving conditional processing as late as possible means that we might process files that don't need to be processed for a specific build target. If we are building multiple targets, however, there is a considerable savings to building the file once.
irondog1970
Just getting started
I'm downloading the Priestley Windows Media file, and I'm going to schedule use of the video projector for this Friday to watch it. I'm hoping that with it and the book I'm requesting to be purchased will be a good start to learning what DITA is and how we can use it. As I explained to Kay Ethier yesterday:
erikh
About: The OASIS Symposium, or worth its salt
"I have met with a philosophical work in which the utility of salt has been made the theme of an eloquent discourse; and many other like things have had a like honour bestowed upon them."
— Plato's as translated by Benjamin Jowett
Most conferences serve a particular community. For instance, the attendees at Content Management Strategies are all on the same wavelength. ("We want reuse!") Not so at the OASIS Symposium, where attendees have a common means (XML standards) for a multitude of goals. The diversity of human endeavor can be a bit mind boggling. NIST naming and design rules, SAML swimlanes, Business Process messaging -- who knew? Which has a less delightful dark side: who could know it all? And without a know-it-all, does this important work end up constructing an acronymic Tower of Babel? Does a standard that solves a specific problem in isolation creates integration problems elsewhere? That question came to the front more than once.