[ImageJ-devel] Auto-Documentation of Processing Chain: Extension of @Parameter Annotation

[Oliver Gress]

Hi Gabriel,

> I had a brief look at the pdf. I am not clear (but this is just me) how the 
> history graph differs from a macro/script recorded sequence.
> The nature of processing (or workflow) is sequential and I guess that from a 
> recorded script a human readable form might be created, no?
You are right, you could reconstruct the history of a result by the
script/macro. But there are some pitfalls like modification of a script
or changes in software version of IJ or some plugins, which you would
need to record manually.
I have to admit that we barely use macros, but write plugins for IJ,
operator classes and standalone java applications.
The history concept was therefore developed to work beyond the level of
plugins to keep track of input parameters, how results were computed,
i.e. which operators were involved, and which versions of the involved
operators were used in a completely automatic way. This history is
firmly associated with a result, while a script/macro is not.

> I presume that implementing something like this would require that all current 
> commands in ImageJ should output something to create the history graph?
Yes that's true. We are not able to record ImageJ-operations. Using
these standard IJ-operations would break the recording of the processing
chain. We currently can only wrap IJ-plugins to be used with our history
concept. A modification of the @Parameter annotation would therefore be
a step towards better integration at least with IJ2.

> You might be interested in http://imageflow.danielsenff.de/
We already had a short look at imageflow. We were thinking about visual
programming using our concept, because every operator's interface can be
inquired. Imageflow has a great visual representation, but as far as I
know needs manual configuration (xml-specifications?). We need to check
if we could create these specifications automatically.


Best regards
Oliver