Subject Re: [IBO] howto: do unicode with IBO and FB 1.5.1
Author Daniel Rail

At August 20, 2004, 17:35, Constantijn Wolfs wrote:

> Daniel Rail wrote:

>>At August 20, 2004, 15:54, mircostange wrote:
>> > Not sure where to start.
>> > I have an app that is based on D5 and runs pretty fine. Now I would
>> > like to move to unicode support.
> I had the same idea some weeks ago!

>> > Perhaps someone can give me a step by step intro how to get there.
> I've recently posted this question in Fb-Support ng and this list
> (twice), without any answer whatsoever.
> It seems that unicode is still not a topic.

>> > - what do I need to do when creating the database (is CharSet
>> > UNICODE_FSS or NONE appropriate)?
>> > - do I need to do anything specific when creating tables?
> COLLATION is lost when you don't store as UNICODE_FSS.

>> > - what client can I use to put some unicode text into the db?
>> > I tried IBSQL, but either it doesn't do the job or I didn't do
>> > it right. What are the steps?
> Free TNT controls are terrific, but in my expirience you loose a lot of
> performance. I use a lot of lookupcombo's in my app and even in ANSI
> performance degraded significantly.
> I read that Mike Lischke and Allan Casteran have donated the
> AsWideString support to Jason. However, as far as I can see it is not
> implemented yet. And also
> it looks to me not a simple thing to do. Take for instance ParamCheck
> which must always be set to false because Delphi will convert to ANSI
> automatically.

>> > I am willing to move to D7 and the latest IBO version if necessairy.
> I decided to wait how Borland picks up unicode seriously.

>> > - what do I need to do in D7/IBO in order to have a form with
>> > editboxes, etc, that contain unicode chars?
> In D7 at least W2K and for instance Arial Unicode MS installed. You
> could turn to ADO
> for a unicode data-access layer. The TNT LX component set contains ADO
> table-, query-
> and dataset-components.

>>Have you read Delphi's help on Unicode Characters? BTW, what's in
>>D5's help is almost the same as in D7. Delphi's VCL still uses single
>>byte characters, and Unicode is multiple bytes.
> Aren't you curious how Borland takes on unicode in D9 or D10 or D11 or ...?

I'll read about it when it's written. But, D8 supports Unicode via
the .Net framework. I would be pleasantly surprised if Borland
changes the Win32 VCL in D9 to properly support Unicode.

>>And, why do you want UNICODE_FSS? If you will only be using European
>>characters(i.e.: accents), then maybe you should look at the ISO8859
>>character sets.
> We don't specifically want UNICODE_FSS (3 bytes? Why?); we want *unicode*.

Unicode is a multi-byte character set(MBCS), which essentially means
one or more bytes per character.

> I read that Chinese is the favorate food among Firebird gurus.

I don't know where you heard/read that! The support of Unicode in IB
and FB natively via UNICODE_FSS(UTF-8) is not the best. That's the
reason why Peter Jacobi is helping in improving the Unicode support in
FB and maybe(it's being talked about) in FB 3.0, the support for
UTF-32 will be in place.

Best regards,
Daniel Rail
Senior System Engineer
ACCRA Group Inc. (
ACCRA Med Software Inc. (