Hi All,

Originally I'm on record as opposing the idea of generating our records
and types via autogeneration from XML definitions.  I now see Glen was
absolutely right.  I was afraid that supporting it would be more trouble
then it was worth and bla bla bla.  Anyhow I was also concerned about
"subclassing" idea Glen brought up but I'm game to it provided its done
in a way thats not confusing.


for instance

org.apache.poi.hssf.record.definition
would contain the autogenerated record stub  say BOFRecordStub

org.apache.poi.hssf.record
would contain the stub subclass that had the meaty functions in it.

the definitions should contain at least one abstract function that
prevents them from being instantiated or perhaps protected constructors
or something.  Subclasses must clearly point to their definition and
vice versa.  The initial subclass could be autogenerated for
implementation speed, but should be deliberately non-working.

We might consider doing this for hdf of course as well.  

This would allow us to keep the autogenerated records up to day and in
synch with the xml definitions w/o breaking concrete complex
implementation functions.  I also think it minimizes the complexity and
might even be reasonably sound oopwise because it seperated
non-structural concerns from structural concerns (meaning the binary
structure and functions from the stuff we have to do to stuff data in
and out of it)...  

Thoughts, comments, social criticisms?

Should we convert the old stuff...  Coming up with a method to do that
sounds like a job for Super-NicolaKen.

Anyhow, me thinks I have the starts of a good run for HDF.  I'll shortly
begin banging out HDF definitions.  I hope to bang out FIB this weekend.

Hopefully this will scare Ryan into coming out of the dark and work in
the light..  :-)

-Andy

-- 
http://www.superlinksoftware.com
http://jakarta.apache.org/poi - port of Excel/Word/OLE 2 Compound
Document 
                            format to java
http://developer.java.sun.com/developer/bugParade/bugs/4487555.html 
                        - fix java generics!
The avalanche has already started. It is too late for the pebbles to
vote.
-Ambassador Kosh

Reply via email to