V4RB 1.11 - serious bug in index by words

[erne]

Hello Ruslan,

I follow up with some more findings

On 13 feb 2005, at 16:21, erne wrote:

> Hi Ruslan,
>
> On 13 feb 2005, at 09:11, Ruslan Zasukhin wrote:
>
>> On 2/13/05 12:07 AM, "erne" <ernestogiannotta at tiscalinet.it> wrote:
>>
>>> Hello Ruslan,
>>>
>>> I've found that when a field is Indexed By Words the high ASCII
>>> characters are not found
>>>
>>> e.g.
>>> FieldIndexedByWords.SetString 'ni=96o'
>>> c =3D db.SQLSelect('Select * from Table Where FieldIndexedByWords =3D
>>> 'ni=96o')
>>> // returns no finds, this is true for every accented character!!
>>>
>>> this is also with 1.10 and a big showstopper to me
>>> can you please take a look at it?
>>
>> Hi Erne,
>>
>> You sure that field have correct language parameter?
>>
>
> well... I double checked and found it a bit more complex
>
> I now can consistently reproduce the bug now
> only in Text method fields (the one I use for all around search)
> and only for certain high ascii chars (like =C3, =BC and many others)
>
> in fact I have not consistent results with String method fields...
> they seem to be found sometimes and not found some other times
>
> language setting seem to not affect...
> I got a sample project & db... can I send it to you?
>
> it's quite  important for me since it breaks lots of searchs in my app
>

ok I got it... seems problem is all these strange characters are=20
considered as separators...

well this breaks unicode indexed by words searches in V.1

...I was asking for too much luck ;^)
(if only we had the ability to define word separators for index by=20
words...:)

OK, I heard it... let's move to V.2 :!)


Cool Runnings,
Erne.
|er| musical box
|ne| a media store