Drew Jensen escribió:
Thought I would put this up while I am at it.
Given Darren's earlier post I decided to create the same basic report
with the wizard. The behavior is quite different.
I started the report at 3:52:39
The report rendered to the screen at 4:05:15 showing page 1 of 12
At this point control returns so that you can work with the system and
the report starts adding pages.
It is done mow 1033 pages, and the time 4:16:04
Now before I go play with postgres I think that it would be better to
create a more accurate report for testing.
My dataset is basic personal information for professional baseball
players born from 1820 - 1984
I grouped on Year of Birth, Month of Birth, Country of Birth
Details Last Name, First Name, Nick Name, Batted ( R / L / B ) Threw ( R
/ L B )
One thing is that the two tests are not equal.
The report wizard creates a select statement just for the fields used, a
better use of internal buffers for the DB engine, the SRB I don't know
about - but I think it does a SELECT * when a table is used. So I would
also need to compensate for that by using a Query in both cases and
would create indexes on the fields that would be sorted in query before
handing it to the report engines.
Would we do this in the real world, well not often to get a 4224 page
report like the first or even the 1033 report just now. Once maybe, so
is 16 minutes acceptable for a report that in the real world is most
likely a one time run?
Well, this brings me to my question in the first mail: is a 20.000
records table a "normal" test case for this kind of report builder?
I still didn't get the answer about what exactly the user is trying to
do. If he wants to "dump" the whole table to paper, the report isn't a
very good idea, I'd rather try using css.sdb/c/x together css.ucb and
css.io to write to a stream (well, no formating and other pretty stuff,
but...).
A more interesting test would be to get some stats:
I'll think of how I would I would slice and dice that data and then
create the two different approaches.
Case one - Generate the statistics in SQL statements and use the wizard
to display results
Case two - Generate the stats. using functions in SRB ( since Ocke was
good enough to render the wiki pages, might as well put that information
to use )
I'll get that put together and some metrics on both approaches tonight.
You know this is actually very good - everyone ( ok, me at least ) has
been so taken with the NEW, that we almost forgot there was an OLD that
still has uses.
Yes, and it seems that they can live together (I hope the wizard does
not disappear in future releases), I'm not completely sure at this time
if the SRB replaces the wizard, or is there for the user that wants a
more complete and complex report, while if you want a simple table-like
report you keep using the wiz.
Finally, let me point something very positive concerning OOo, and Base
in particular: if you read the user's mail, you would have notice that
his e-mail ends with @villaelisa.gov.ar, this means:
* he is from Argentina, just like me
* he is working for the Government in Villa Elisa (Córdoba, Argentina)
The important conclusion here is that the Government of Villa Elisa is
working with OOo, instead of MSOffice, and in particular with Base.
As Argentinian, this have surprised me a lot; let me tell you that OOo
developers, and in particular Base developers, should take this as some
big achievement of their part: Microsoft has here a lot of power, just
like everywhere, but here politicians are cheaper!
Bye and luck!
--
Ariel Constenla-Haile
La Plata, Argentina
[EMAIL PROTECTED]
[EMAIL PROTECTED]
http://www.arielconstenlahaile.com.ar/ooo/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
