Re: Patch for the handling of symbols in clsql

Howdy CLSQL List,

I made a first draft at fixing the symbols issue we had been 
discussing.  While tackling this I also knocked out the 2 TODOs in 
read-sql-value.  I also introduced a specific error type 
'sql-value-conversion-error instead of that read-sql-value generating 
many, different, generic errors.

https://github.com/UnwashedMeme/clsql/commit/oodml-symbol-values-refactor

I had been wrong in previous emails, with some of my concerns about 
conflation of symbol-typed view-slots.  These are sent to the database 
via database-output-sql-as-type instead of database-output-sql, and so 
have an opportunity to print this correctly before dropping into the 
fdml.  As such my previous assertions of output and symmetry between 
reading and writing were also wrong (as I had looked at the wrong function).

I have recently moved to a new computer and don't quite have all my 
clsql testing stuff worked out.  I still need to get higher order test 
objects defined (eg: view-classes with symbol-typed slots), write tests 
for those, and actually get my tests running.  However I figured I might 
go ahead and put a tentative patch out there for those waiting or 
willing to assist in testing.

Cheers and thanks for the feedback,
Russ Tyndall
Acceleration.net