[Libreoffice-qa] db test VM (was: automated testing of Base)

drew drew at baseanswers.com
Fri Oct 7 10:28:15 PDT 2011

On Thu, 2011-10-06 at 17:13 +0200, Alexander Thurgood wrote:
> Hi Terry,
> Le 06/10/11 16:40, Terrence Enger a écrit :
> I can only speak from my perspective of an "enterprise" user, and thus
> possibly biased ;-)
> > Do we have a feeling for what databases are most used "for real" with
> > Base? Or how they are used? 
> ODBC is still used as an enterprise-wide solution in many places because
> of its generally good performance, usually much better than the
> corresponding JDBC drivers especially with regard to OOo/LibO. The whole
> Java encapsulation thing always had performance issues that ODBC simply
> did not. That being said, there seem to be far more actively maintained
> JDBC drivers around these days than ODBC drivers, probably because it
> fell under the "write once/run anywhere" (uh-huh) motto that was touted
> for years with Java development.
> > Or what open bugs are most deserving of attention? 
> Too many to be reasonably approached by one person in a lifetime ;-) 
> Just kidding, but there are quite a few, as with the other modules, and
> depending on who you ask, they _all_ deserve attention (grin).
> > How much real use is through ODBC as compared to JDBC as compared to
> > built-in drivers? 
> Hmm, like I said, from what I have seen being used in various companies,
> JDBC would appear to come out tops in usage terms because of its
> platform/OS agnosticity.
> The built-in "native" ones are however highly appreciated because they
> appear seamless and don't depend on Java...and on the whole their
> performance rocks, whereas JDBC performance within LO dives rapidly,
> depending on the release version of Java, the driver itself, and the
> poor memory management within Base where everything has to be loaded and
> held in memory until you write out it all back out in the IOstream to
> the file.
> I had always implemented ODBC enterprise wide in earlier versions of OOo
> on Linux because (1) there were no native drivers at the time, and (2)
> performance was excellent with ODBC, but many ODBC solutions for some db
> engines are only available for one or two OSes (for example on Mac, it
> is a rather large PITA no thanks to Apple), which obviously is a serious
> slug in the chest when you have heterogenous environments. That is why
> the native cross-platform driver is so appealing :-)))))
> Obviously, all of the above just represents my small window on the world
> of DB connectivity that I've either experienced, implemented or come
> across in the last 20 years.

Howdy Alex, Terry, et al

OK - well, how about I change the subject line but not the thrust of the

I read over some material regarding the opensuse.openQA goings on, and
watched a video file about same...so, I'm quite dumb still with specific
regards to the openQA scripting environment/tools still but if I look at
the idea of a test suite for connectivity and Base module functions that
I could put to use, for myslef, right now then I figure the SuseStudio
server might be useful. Plus I kind of figure there must be some kind of
relationship between the tool chains..maybe.

Alright - so I did a little fiddling last night and this morning and
started putting together a test best image:

So what I'm thinking to do with this image definition is:

1 - a common, known, environment:
[at the moment this includes]
-- Address book sources for mail-merge testing 
--- Evolution
--- Thunderbird

-- File based data sources
--- SQLite 3
--- dBase
--- embedded ODB files

-- Server based data sources
--- MySQL
--- Postgres

2 - pre-configured data sources
--- MySQL and Postgres dump files w/schema and test data
--- SQLite db file and UnixODBC config.ini file
--- Evolution and Thunderbird address book test data
[_not_ finished]

3 - pre-built ODB files connecting to data sources
[_not_ finished]

4 - test scripts

actually, at this point I'll copy that to a wiki page and use that as a
better way to discuss what is going into this as I go

*chuckle*...alright that likely just causes head scratching..but I'll
work on some test data today, update the image definition and do another
build and the wiki page as I do - hopefully doing a better job of
explaining what I'm thinking ... along with thoughts/questions I have
about test scripts, perhaps this can lead to something that would work
in the OpenQA environ, but it not that's ok..this is an itch that I;ve
had and time to scratch...

so - took the thread a bit OT perhaps :-/


More information about the Libreoffice-qa mailing list