Import data - problems

[Erne]

Il giorno 25/ago/09, alle ore 15:11, Thorsten Hohage ha scritto:

> Hi,
>
> On 2009-08-25, at 15:05, Greg Olson-Hyde wrote:
>
>> On 25/08/2009, at 7:38 PM, Ruslan Zasukhin wrote:
>>
>>> Better avoid such transformations of types in DB..
>>> Just from paranoid point of view :-)
>>
>> Are you just paranoid, or speaking from experience?
>>
>> What could the problems be? I will only be importing once.
>>
>> I have no experience with RegEx and have enough new stuff to learn at
>> the moment!!!!! It would need to get the quotes from "n", "nn",  
>> "nnn",
>> "nnnn" etc but not "Born in 1940" - beyond me right now.
>
>
> As Bart already mentioned, if this is a serious task - and it seems
> too - then *I* would import everything in varchar fields and then use
> one / some SP(s) to move the data to the final destination. Perhaps
> this strategy is easier then to deal with a lot of regex before
> importing.
>
> Btw. many dumps, e.g. one generated withe the _loved_ FileMaker Pro
> can have a lot of invisible chars in columns that are intentionally
> not thought to be used with such chars ->>> many trouble. All this can
> be avoided with such a two step import, too.
>

And I can confirm this, having just been hit by one of those gremlins  
chars in a file exported by FileMaker
(to be fair it was a dbf file coming from Foxbase/mac, opened in FM,  
exported as text and then imported into Valentina)

While we're at this Ruslan, may I ask why the defaults for  
VCursor.import are "," as delimiter and UTF-16 as encoding?
(I'd expect tab and UTF-8, but maybe it's just me... :-)



Cool Runnings,
Erne.