Re: For Windows platform: which Dia plugin or App to simplify design and documentation of small Database Applications ?

[Thomas Harding <tom thomas-harding name>]

Le 28/10/2012 12:38, Thomas Harding a écrit :
> Le 28/10/2012 11:13, Thomas Harding a écrit :
> > I can author a Dia sample, then test it and send both diagram and 
> > result. I think it will took
> > half an hour more (I'm puzzled by GUIs: I learned to read and write, 
> > not to draw)
> > :)
> >
> > And, yes, things has changed since I used for dia2sql :) 
>
> Done *a minimal one*.
>
> [see quoted post for attachments]
Dia can helps on design/design to code, while it is not the only tool 
(especially for
Java/MS-Windows you seems to target on... there is Eclipse)

Dia will help you if you need for separate tools (one tool, one function).
dia2code is a good one (for skeleton).


About other question (which RDBMS/DB engine to use) :

It could not be the "right" question!

http://en.wikipedia.org/wiki/PostgreSQL (really powerful & extensible, 
high scalability)

http://en.wikipedia.org/wiki/HSQLDB (I'm not aware of extensibility (eg: 
new types), can have small footprint, while needs Java: are you aware of 
the 2GB limit/32bits?)

http://en.wikipedia.org/wiki/SQLite (library, really small footprint, 
but no type coercition -- eg as in Python, more likely as in Perl (you 
fix a preffered type, it can change ; while in python a variable is not 
type-ed, but while "fixed" needs ~type cast), also: don't expect 
concurrency, eg to use on a (dynamic) web server this is definitely not 
suitable)

http://en.wikipedia.org/wiki/Comparison_of_relational_database_management_systems

there are also numerous other database models than object-relational 
(and even for last there is not only SQL to query/define):
http://en.wikipedia.org/wiki/Database

So, the real question is: what do you want to do? how big is your data? 
do you need for foreign keys, table inheritance, ...? have you special 
types which needs a real coercition more than to be checked by triggers 
(or externally by your appliance)? is your appliance mono or multi-user? 
is your data relational (eg, see Apache CouchDB)?

In fact, to design your appliance on a schematic (UML class diagram is 
one of, especially to design underlying database, but this is not the 
only and you definitely must not forget the other UML ones!) will help 
you to decide.

Whatever you decide, the benchmarks are your matter and you'll need to 
experiment for your case with several prototypes /and/ a "fake" data set 
suiting your requirements, as possible as large as you'll fill.

My opinion is: excepting for "big data", it matters now only for 
embedded and handhelds (if it fails elsewhere, your code is definitely 
badly designed).

Note: on Linux you have memstat if you want check for memory 
consumption... I expect a graphical tool on MS-W$ also exists.

Regards,
TSFH.