[SqueakDBX] OpenDBX vs ODBC / MsSQLServer

Tue Jul 7 03:37:42 UTC 2009

Hi! What an excellent question. However, I have to admit I never used ODBC
squeak driver.

I hope you wanted to said SqueakDBX instead of OpenDBX :)  I guess you are a
Smalltalk programmer. Actually, we (SqueakDBX) has advantages over other
drivers and also, OpenDBX has over other libraries.

1) First, I will copy paste a piece of text from OpenDBX webpage:
"Contrary to interfaces like ODBC, the OpenDBX library favors speed and
flexibility and therefore doesn't try to unify the SQL query language. It
enables the application developer either to use only standardized SQL
elements or to utilize more efficient database specific language elements as
well. Furthermore, this approach allows adding support for databases which
don't implement SQL as query language. "

2) OpenDBX supports: Firebird, Interbase, MS SQL Server, MySQL, Oracle,
PostgreSQL, Sqlite, Sybase ASE and ODBC. As you can see, OpenDBX has ALSO a
ODBC backend.

3) OpenDBX compiles in Linux, Windows and Mac (not only windows)

4) With OpenDBX you don't need (except you use the ODBC backend) external
configuration as you need when using ODBC (datasources and all that stuff).

5) This is one of the most important: SqueakDBX, as ODBC, uses FFI to
communicate to external functions. As you may know when you call a function
with FFI, the COMPLETE VM is locked until the function returns. Suppose that
function is the one that process a query. And suppose you have a
WebApplication running in your Squeak. You will lock ALL the VM to ALL users
until the query is done. This can be very very bad.
OpenDBX (when possible) uses asynchronous queries and has a timeout to retry
the query to see if it is done. But this wait is in Squeak, not in C. So, if
there are other request waiting, the scheduler will process them. So, the
time the VM is lock is shorter than in ODBC.

6)  I don't why you want to use ODBC but do you want to write SQL by hand?
We don't :)
Glorp is an ORM for Smalltalk. This is actually for VW and unfortunately,
the Squeak port only works on PostgreSQL. We are modifying it and
integrating Glorp with SqueakDBX. So, you will be able to use an ORM for all
backends. For more information, read:
http://www.esug.org/Promotion/SummerTalk
We hope to have this integration by the end of the year. Right now we have a
proof of concept working with Glorp and SqueakDBX for PostgreSQL :)

7) We have 90 unit tests that are run for every backend.

8) Lots of benchmarks and comparison with native drivers (PostgreSQL and
MySQL). We don't have benchmarks comparing to ODBC but I would love to have.
If you do something, please send your results.

9) SqueakDBX is very well documented. You can see our wiki page:
http://wiki.squeak.org/squeak/6052
In addition, we are creating our own seaside webpage, but it is not ready
right now.

10) We have several features like:

   - Transactional concept: begin transaction, commit and rollback
   - Mappings from String to specific types in selects
   - Special OpenDBX options: multistatements, encryption, compression,
   paged results, mySQL modes, and more
   - Automated database connection release (although manual disconnection is
   recommended ;-)
   - Automated results retrieving in order to do another query, after doing
   a query and not iterating ALL results
   - Error handling: Not only errors, but levels associated with an error in
   order to avoid FFI calls (if you get a fatal error, it has no sense to do
   another query and the resources must be free).
   - Query timeout (this is very useful for webapplications) and pageSize
   can be set for each query.

Obviously, we have our drawbacks: We don't support large objects (blob,
clob, etc..) yet neither Stored Procedures or functions (altought we support
multistatements). We also have little "production" uses of SqueakDBX, as it
is relative new.

You other questions:

Yes it works perfectly well in Pharo. Actually, half of the team uses Pharo
for the development and the other half, Squeak. We are also the one that
makes Glorp to work on Pharo. See my email in Pharo list:
http://n2.nabble.com/-ANN--Glorp-now-load-easily-in-Pharo-td3056236.html#a3056236

You see an example, I would read the getting started in here:
http://wiki.squeak.org/squeak/6061  All the examples there work for all
backends (they work for mssql). This is the interesting things: the code is
exatly the same to all backends. The only thing it could change is the sql
written (if you wrote vendor specific sql).

That's all I imagine right now.

If you have further questions, please ask.

Best,

Mariano

On Mon, Jul 6, 2009 at 7:17 PM, Claudio Campos
<claudio.sistemica at gmail.com>wrote:

> Dear Mariano,
>
> What are the advantages of OpenDBX vs ODBC?  How is the integration with
> Pharo? Glorp? Are there examples for ms sql server ?
>
> Thanks in advance,
>
> Claudio Campos
>
>
>
>
>
>
> _______________________________________________
> SqueakDBX mailing list
> SqueakDBX at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/mailman/listinfo/squeakdbx
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeakdbx/attachments/20090707/cb0f1907/attachment.htm