Hi Emerson,

I hesitate a bit as I have no clear idea what your priorities are.
Compared to PostgreSQL codebase is a lightweight,
but that is not neccessarily a bad thing from admin POV, ease of install, resources and so on.

My "dream solution" is a cursor engine running on Intel and Pads/Phones with ARM, 515MB Ram and enough
Flash for cursors / local tables. Having static lookup tables installed locally and able to join with the server side data
IMO is the best app architecture - and ORM heavyweight baggage not always helpful.

So I can visualize some architectures where a lean library like Sequitur might fit in.
I am certain that FP-fields are handled ok, but have grown accustomed to datetime, int and currency -
check out if Sequitur supports those as of now. Not supporting binary index is not that bad if they have C/S.

Perhaps Advantage might be a good compromize between Sequitur and PostgreSQL.
For me - if Lianja does not grow into something sellable - the idea was to create an intermediate layer in
Python or Java utilizing SQLite, as SQLite is available already on all platforms, or one Java source embedded DB.

If you are considering mobile and want a cursor engine, look at Lianja - certainly not a safe bet currently,
but very interesting and integrates a vfp-Like xbase implementation plus optional DB C/S engine
with JS, Python, PHP and a framework runable across many devices.

>Compatibility doesn't means that I can get the best that CodeBase can offer using the VFP native xBase commands to access this data. To get the best that CodeBase can offer, I think I'll need to access this data using an ODBC driver or the CodeBase API.

Correct if you consider WAN access - real server processing will beat even optimized vfp-ISAM in many scenarios.

>If I'm right, the work to change our apps to use CodeBase would be the same of change to use any other database. So, I prefer to use a database like PostgreSQL, for example, that is free and will do the same that CodeBase does.
>Do you agree?

If you plan to keep the clients in vfp for long - yes. Getting the DL exchangeable would be IMO the best strategic move -
have a cleanly architected solution where you CAN be nimble faster. And PostgreSQL is certainly a valid backend.

But if your future (but not too far) plans include something like cursors for clients, there are possibilities.
Having one clear and maintainable backend for server up to local engine eliminates some problem areas.
Certainly a benefit / plus for the offerings of Codebase, Advantage and Lianja.

sorry to be that unclear/undecisive

thomas