Vincent,
Your ideas a sound, but are probably more suited to a new Application model
and with the Neo release would there be any point? Surely when the new
Server tech is released, the best bet would be to make Spectra portable,
free and more reliable....
In my previous threads and IMHO, the strength of Spectra and the COAPI was
intended to be "the API itself" rather than the object persistence
architecture. Although having a consistent database schema between all
spectra projects makes a user's SQL code somewhat portable, I would assume
the original intent was that they didn't have to write SQL code at all!!!
As for speed; I would assume many spectra users are writing their own SQL
because the queries to access content objects in the Spectra schema are
simple to infer and the custom written SQL performs better than the generic
queries of the API. Maybe a whole re-address of this area is in order.
However, this approach was dangerous for one major reason: users' custom SQL
was not guaranteed to portable between different versions of spectra. If the
schema changed between say version 1.0 to 1.5 (which it did) then that SQL
code may have to be re-written.
However, an API abstracts the underlying queries and protects the user
between versions of the schema. Also, an API is much more portable between
projects.
IMHO, Spectra's flat, serialized data architecture made object look-up by ID
fast (however ugly), legacy type definitions less sensitive (although an
ALTER table command in a traditional relational model is still pretty easy
to do) to property changes etc... BUT it still missed out on a lot of
benefits of a relational design such as "default property searchablility",
strong referential integrity, reduced redundancy etc... I strongly believe a
relational model going forward will be fantastic. I have heard that a major
focus of Greylock's COAPI is to address the concerns people had with the
Spectra COAPI. A major one will be to ensure that they don't have to write
their own SQL against a datasource. The API will do it all - the API will
take into consideration more than just "performance concerns" for why people
wrote their own queries. Users will still want to do conditionals and their
own sorting etc, but users should forget about directly querying the
relational model.
As I said, Spectra needs to be a portable application which is free, does
not rely on any 3rd party prods for functionality other then what is
available to the public and not just thru MM. The product has been given
its death date and after that who knows? It aint gonna go away as a CF app,
and thats what a lot of poeple I think will look from it. It is up to us as
Spectra developers to make sure that it is sound. :-) Oh, lose
SiteMinder...please :-)
Neil
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Your ad could be here. Monies from ads go to support these lists and provide more
resources for the community. http://www.fusionauthority.com/ads.cfm
------------------------------------------------------------------------------
To Unsubscribe visit
http://www.houseoffusion.com/index.cfm?sidebar=lists&body=lists/spectra_talk or send a
message to [EMAIL PROTECTED] with 'unsubscribe' in the body.
