FreeTDS Development Group

[Frediano Ziglio <freddyz77 AT tin.it>]

> I have (pruned for mail):
> 
> 1> sp_configure
> 2> go
>  name                        config_value run_value
>  --------------------------- ------------ -----------
>  default full-text language          1033        1033
>  default language                       0           0
>  show advanced options                  1           1
>  user connections                       0           0
>  user options                           0           0
> 

Same

> 1> sp_helpsort
> 2> go
>  Server default collation
>  --------------------------------------
>  Latin1-General, binary sort
> 
> > All this stuff are defined in windows...
> > 10 bit for language (low), other for sublanguage. 
> > English (9) american (1) is 0x409 -> 1033
> 

Mine is 

Server default collation        
Latin1-General, case-insensitive, accent-sensitive,
kanatype-insensitive, width-insensitive for Unicode Data, SQL Server
Sort Order 52 on Code Page 1252 for non-Unicode Data    
(return status = 0)


> Yes, but:
> 
> 1> select langid, lcid, msglangid, cast(name as char(30)) as name from
> syslanguages where langid = 23
> 2> go
>  langid lcid        msglangid name
>  ------ ----------- --------- ------------------------------
>      23        1033      1033 British
> 
> The "sublanguage" wouldn't affect collation.  Maybe that's why it doesn't
> appear in syslanguages.  
> 
> > > > - 2 bytes ???
> > > 
> > > My next two bytes are always 00 01:
> > > 
> > 
> > Mines are 0d 00 (0xd0 in LE??)...
> > Full lcid on windows as also sort (bits 19-16) and version 
> > (bits 23-20)
> 

Perhaps 
Mine 0xd0 -> 0000 0000 1101 0000 -> third: 1 width insensitive 1
katatype insensitive 0 accent sensitive 1 case insensitive
Yours 0x100 -> 0000 0001 0000 0000 -> second: 0 0 0 1 binary third 0000
all insensitive.
Other bits ??

> Let's say we don't know anything about the next three bytes. 
> 
> I have:    00 01 00
> You have:  d0 00 34 
> 

Last is 0 or SQL charsets id (only used for SQL coding)

> Our servers are configured the same way, except I use binary sorting order
> (33280, 0x8200).  
> 
> If the fifth byte is, as you said, syscharsets.id, then mine should be 0x32
> (50).  But mine's the default  <shrug>.  Perhaps I'm using SQL collation
> instead of Windows collation?  (I don't know how it was set up.)  I can't
> reconfigure my server to test that theory.  
> 
> Your 0x34 tells us the sort order (and implies the character set).  
> 
> That leaves bytes 3 and 4.  If they're a 16-bit int, you have 0x0d (13) and
> I 0x100 (256).  
> 
> It can't be our client settings, because we don't transmit character set
> information to the server, to it can't know what we use/want.   
> 
> So far, the server's told us about the column:
> 
> datatype, size, name 
> locale id (bytes 1-2)
> syscharset.id (collation method and character set, byte 5)
> 
> What more could we want to know?  Country, language?  
> 
> Do we care?  
> 
> Regards, 
> 

Are all supposition...

Yes we case, I must send something to the server. I just fixed this. For now 
i sent same data passed from server during login (handling env change type 7)

Also I exec this query on a mssql2k 

declare mycur cursor for select name from ::fn_helpcollations()
declare @name varchar(200)
open mycur
fetch next from mycur into @name
WHILE @@FETCH_STATUS = 0
BEGIN
        EXEC('SELECT ''pippo'' COLLATE ' + @name + ', ''' + @name + '''')
        fetch next from mycur into @name
END
close mycur
deallocate mycur

Results at http://freddy77.tripod.com/downloads/out2.txt.bz2 (I used bz2 for 
size, now is 97kb compressed, not suitable for mail)
Still unwatched...  

freddy77