Obviously, the Palm OS does not lend itself to multiple running processes or multi-threading. We were thinking more along the line of shared libraries. In other words, when the renderer calls the method for accessing bytes in a document, that method would go directly to our implementation code which would act accordingly (such as parsing the XML, returning the results, etc). Of course, there are a lot of details to work out, but if it were implemented in this way using shared libraries (more of a "plug-in" approach), then for the standard Plucker viewer (which needs to be as fast as possible), you could create a shared library "plugin" optimized for Plucker and its DB format... I'm thinking this wouldn't add too much overhead? A little, of course, but if done right, it should be possible to keep it below the perception radar.
Back to the forms... Since, for our needs, we wouldn't need to have forms span hotsyncs (they would all be handled in real time on the device), it seems the more difficult part could potentially be the form controls. I know the Palm API has limited support for dynamically creating controls, but how far back into previous OS versions does this support go? Can these controls be easily embedded into a scrolling document? Or, would it require development of "custom" controls?
Michael Nordström wrote:
On Mon, Apr 01, 2002, Andy DePue wrote:it appears Plucker has support for images, color, forms,
Not forms.However, it also seems that the code for reading/manipulating the
"document" (or Palm DB) is written directly into the rendering code.
Well, there is some "abstraction".Our idea would be to abstract the document access out of the viewer
engine. The renderer would go through an API to access
documents/images/etc. Anyone could "plug in" an implementation of this
API.s
Sounds OK.We would then be able to write an implementation that could
directly pump the output of our XML script to the renderer and give the
renderer direct access to our image data.
How do you intend to run these programs at the "same" time? What
about the speed when you have to access the data through several
extra layers everytime?
/Mike
