firebird-net-provider

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[Peter J. <pj...@wa...>]

Hi Alessandro, All,

"Alessandro Petrelli" <pet...@in...> wrote:
> I have an unicode_fss db (I need to develop a Japanese application) and =
an
> asp.net DataGrid. All fields are shown correctly except those for which =
I
> call declarative databinding, for example:
[...]
> The field OUT_CHANNEL is return from a stored procedure and is a char(1)=
. I
> should expect values like 'P' or 'E', but instead they are returned as '=
P '
> or 'E  '.

The sad fact is, that this field will also happily hold 'PPP' and 'EEE' or=
 
'P=D6' but not '=D6=D6', if you see what is going on here. (Perhaps the NE=
T 
provider actually manages to forbid these INSERTs, but with the C API or 
ISQL it's possible.)

Now, at SELECT time, a clever middle layer (like the NET provider ;-) ), 
can try to make the situation more sane. Knowing that a 3 byte buffer for 
UNICODE_FSS should better only hold one character, it can algorithmically 
extract just one UTF-8 character from the buffer, ignoring the junk bytes 
after that.

Regards,
Peter Jacobi

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[<car...@te...>]

Hello:

> Now, at SELECT time, a clever middle layer (like the NET provider ;-) ), 
> can try to make the situation more sane. Knowing that a 3 byte buffer for 
> UNICODE_FSS should better only hold one character, it can algorithmically 
> extract just one UTF-8 character from the buffer, ignoring the junk bytes 
> after that.

Yes :) the only problem is how to deal with querys against system tables 
( maybe i will be checking that the field name and the table name starts 
with RDB$ ).




--
Best Regards

Carlos Guzmán Álvarez
Vigo-Spain

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[Peter J. <pj...@wa...>]

Hi Carlos, All,

> the only problem is how to deal with querys against system tables (
> maybe i will be checking that the field name and the table name starts with
> RDB$ ).

I've already fiven my disclaimer that I've lost (if I ever had) the 
overview. But...

Aren't the system table columns the only UNICODE_FSS columns with length in 
(in bytes) which isn't divisable by three? 

Thinking of it, this test isn't really better than testing for RDB$, only 
want to bring it up.

Regards,
Peter Jacobi

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[<car...@te...>]

Hello:

> Aren't the system table columns the only UNICODE_FSS columns with length in 
> (in bytes) which isn't divisable by three? 

Known by me yes :)

> Thinking of it, this test isn't really better than testing for RDB$, only 
> want to bring it up.

I'm sorry, i don't understand this, can you explain in another way, 
please :)




--
Best regards

Carlos Guzmán Álvarez
Vigo-Spain

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[Peter J. <pj...@wa...>]

Hi Carlos,

> > Aren't the system table columns the only UNICODE_FSS columns with length
> > in (in bytes) which isn't divisable by three? 
> 
> Known by me yes :)
> 
> > Thinking of it, this test isn't really better than testing for RDB$, only
> > want to bring it up.
> 
> I'm sorry, i don't understand this, can you explain in another way, 
> please :)

The rather hackish idea is:

// Incoming data: n bytes of UNICODE_FSS
If (n % 3 == 0) {
  // extract n/3 UTF-8 characters into destination buffer
} else {
  // copy n bytes into destination buffer
}

Regards,
Peter Jacobi

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[<car...@te...>]

Hello:

> // Incoming data: n bytes of UNICODE_FSS
> If (n % 3 == 0) {
>   // extract n/3 UTF-8 characters into destination buffer
> } else {
>   // copy n bytes into destination buffer
> }

Huummm i will test it later today and post the result, thanks very much :)




--
Best regards

Carlos Guzmán Álvarez
Vigo-Spain

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[Alessandro P. <pet...@in...>]

"Carlos Guzmán Álvarez" <car...@te...> ha scritto nel messaggio
news:403...@te......
>
> Huummm i will test it later today and post the result, thanks very much :)
>

I tested latest cvs source, now varchar are correctly retrieved but char
no... any idea?

Ciao,
Alessandro Petrelli.

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[<car...@te...>]

Hello:

> I tested latest cvs source, now varchar are correctly retrieved but char
> no... any idea?

Because now varchar fields are readed/written trimmed, but char fields 
no ;) , i have implemented the solution proposed by Peter Jacobi only in 
1.6 sources in my local tree.





--
Best regards

Carlos Guzmán Álvarez
Vigo-Spain

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[<car...@te...>]

Hello:

> I tested latest cvs source, now varchar are correctly retrieved but char
> no... any idea?

Because now varchar fields are readed/written trimmed, but char fields 
no ;) , i have implemented the solution proposed by Peter Jacobi only in 
1.6 sources in my local tree.





--
Best regards

Carlos Guzmán Álvarez
Vigo-Spain

Re: [Firebird-net-provider] Unicode and Databinder.Eval

[Alessandro P. <pet...@in...>]

""Peter Jacobi"" <pj...@wa...> ha scritto nel messaggio
news:403E73F2.11159.130DA2D@localhost...

> The sad fact is, that this field will also happily hold 'PPP' and 'EEE'
or=
>
> 'P=D6' but not '=D6=D6', if you see what is going on here. (Perhaps the
NE=
> T
> provider actually manages to forbid these INSERTs, but with the C API or
> ISQL it's possible.)
>

Yes, you're right, all char fields are not trimmed out...

Ciao,
Alessandro Petrelli.