+0 still as-is.
+1, if Jeremias will let me melt his PageViewportProvider directly into
PSLM; it's only about 75 extra LOC and four methods so I think we can go
without the extra moving part.
+1000, if Jeremias will let me remove his RendererFactory.
IMContinuedHO, whatever its elegance, it is not mathematically/logically
reduced and hence unattractive to the Don Knuth types we need to get on
this team, and to the university types/research organizations I would
like to get more interested into this project.
The Don Knuth types do not like to see this:
x + 3 = y + 2 (math)
or (comp. sci.):
i = i + 1;
i = i + 3;
They wrinkle their noses and run away. Math/algorithm-lovers very
rarely derive pleasure from non-fully-reduced work.
They do like to see this, however:
x = y - 1 (math)
i = i + 4; (comp. sci.)
Here, now we're talking algorithmic attainment, their field, and they
are more likely to help us out in our work, FOP being considerably more
interesting than the types of stuff they normally get to work on.
I believe XSL is more than just implementable, it is actually
*solvable*--i.e., multiple teams working independently would arrive at
the same common, timeless algorithm if they all reduced their code while
working on it (just as they, working independently, would all arrive at
the same Pythagorean Theorem.) I am unsure, but I believe it is the
determination of this common algorithm that would be most likely to
attract support from external organizations. RendererFactory, like most
Java design patters, appears to move us away from the common algorithm.
w/o RF: (fully reduced, every team will sooner or later arrive at):
w/RF: After (not reduced, each team could implement differently, per
the tastes of individual developers or groups of developers):
FOP's greatest allure is its unique math-based algorithms, so unlike
most (boring) Java apps you do not need to pile on Java design patterns
to make it appealing to would-be committers. Furthermore, such moves
actually tend to have a negative effect, as it just makes FOP look like
any of thousands of other open-source Java applications (that we would
now need to compete with for resources). In terms of attracting
developers/outside companies, it seems to me that it would be like
putting a magazine cover girl model in an Eskimo outfit, or like taking
a perfectly prepared steak and pouring that horrid A-1 steak sauce over it.
Vincent Hennebert escribió:
Sorry to insist, but what is your final decision? A quick summary of
the thread gives:
- Jeremias +1
- Simon +1
- Glen +0
The FOray+aXSL stuff would be used in the form of a jar file put in
the lib directory; FOray's avalon loggers would be wrapped into a JCL
adapter for now.