Basic questions

[Richard Altenburg]

On 10-05-2003 04:16, "Kem Tekinay" <ktekinay at mactechnologies.com> wrote:

As a 4D user for more than 10 years, and a Valentina user for only 2 months,
maybe you should ignore my advice :-p

> The benchmarks page hasn't been updated in a while. Do the numbers still
> hold true with 4D 2003? Have more recent tests been performed?

Those benchmarks might still be true, even though I haven't tried 4D 2003
myself (still using 6.7). But not everything is about raw read and write
speed. At our company, 50 users are logged on to our 4D Server, and are
running our wholesale job by inputting orders, asking for statistics,
maintaining production schedules, and so forth. 4D is well equipped to this
task. Our database isn't that big, only 3 GB, with a couple of million
records in some tables, and I only think it performs a little worse when
doing big queries and sorts on a few million statistical records. That kind
of job would be great for Valentina.

> 
> STRUCTURE
> =========
> In 4D, if I want to change a field or add a field, I just do it. Even if I
> later open an older data file, it is updated to match the structure
> automatically. Is this true for Valentina data files?

Nope, 4D is a completely integrated database management system with
development environment. That makes these kind of changes very easy to do.
Also, the 4D language is easy to understand and completely targeted at doing
database related tasks, whereas REALbasic is far more suited for all things
NOT related to databases. You can make RB tightly integrate with Valentina,
and changing your database structure is also possible, but you would have to
write some pieces of code to manage this automatically for you.

> When setting up Valentina, you have to choose the number of data files (1
> through 8). Why would I choose 8 vs. 4 vs. 1?

One reason for a separate index-file, for example, is that you can just
throw it away when the index seems to be corrupted. On opening, Valentina
will automatically rebuild it. In 4D, you would have to use the Tools to do
this. Further file separation might be handy, but I think it would be better
if I could tell Valentina where to store the individual parts (for example,
to put the BLOBs on a different volume).
 
> You set the cache size when opening a database. Is there a rule of thumb for
> how this should be set? Is there a point of diminishing returns if it is too
> large?

Paradigma advices you to use half of the user's available RAM for cache. I
experimented with values between 8 and 128 MB, and all is speedy, and I did
not notice a fallback in speed as the cache got bigger. Just like 4D, you
have to be careful to flush the cache frequently, because else you might
loose a lot of data.
 
> Any help will be appreciated. I am really leaning towards 4D for this, but
> would love a good reason to use Valentina.

The lower price of the RB/Valentina combo is a good reason. Also, the speed
of Valentina is great when it comes to millions of records.

But I think you will have to choose between RB and the 4D environment too.
As a database, 4D is great, but its integration with the language, its power
to work with multiple users but also multiple developers, and its proven
technology, are hard to beat. For bigger projects that mainly contain
database related tasks, I would never choose RB/Valentina (especially RB).
But I chose RB/Valentina for my latest development of a single user Internet
binary reader, because it had to do nice interface things (4D is too dull
for that, and not capable enough), and needed speedy searches on plenty of
records (both can do that), and was on a tight budget (RB Standard +
Valentina is hard to beat). When you also need Windows compiles, both can do
the job.

I now make a shareware program, and the license for RB is much better than
4D. The latter wouldn't allow me to write cheap software. But for a single,
large project for one company or something, I would definitely turn to 4D
Server.