On Wed, 10 Dec 2008, t [dot] j [dot] a [dot] devries [..] ... wrote:

> After having used it for some time now, I much appreciate the way in
> which the TaskContext has been designed and structured. It is also
> beautifully depicted in the figure at the bottom of e.g. the RTT
> frontpage.
> Is there a document that details the design rationale behind it? That is,
> a report or so that describes the considerations that have been made, the
> pro's and con's of the resulting design (as you see it), ideas for
> further development, etc., rather than the (also useful!) usage/code
> oriented documentation that can be found on the website.
>

_the_ real reason for its design is not so much technical, but rather
"psychological": after some trials with other programming primitives, we
found this TaskContext concept a very good trade-off between complexity on
the one hand, and understandibility by our typical PhD students on the
other hand. The "complexity" aspect was very much inspired by the concept
of component-based programming, where the CORBA Component Model was the
major source of inspiration; the "understandibility" aspect was driven by
the need to hide distribution and (realtime!) IPC details from our PhD
students. Also maximal platform-independence and maximal decoupling were
major design priorities.

Personally, I am still rather quite happy with the TaskContext as
programming primitive, although we have already a whole roadmap/wishlist of
implementation and standardization issues that we would like to revise for
the 2.0 version...

Herman