[Gambas-devel] Support of LONG in database drivers

[Benoit Minisini]

On Tuesday 15 November 2005 15:27, Andrea Bortolan wrote:
> Hi Benoit,
>  I'm working on it......just a question about the code you've posted:
>   in the con_data function there is a duplicate case for the
> FIELD_TYPE_LONGLONG see the following :
>
>
>
> static void conv_data(char *data, GB_VARIANT_VALUE *val, int type, int
> len)
> {
>   GB_VALUE conv;
>   GB_DATE_SERIAL date;
>   double sec;
>
>   switch (type)
>   {
>     case FIELD_TYPE_TINY:
>
>       if (len == 1)
>       {
>          val->_boolean.type = GB_T_BOOLEAN;
>          /*GB.NumberFromString(GB_NB_READ_INTEGER, data, strlen(data),
> &conv);*/
>          val->_boolean.value = atoi(data) != 0;
>       }
>       else
>       {
>         GB.NumberFromString(GB_NB_READ_INTEGER, data, strlen(data),
> &conv);
>
>         val->_integer.type = GB_T_INTEGER;
>         val->_integer.value = ((GB_INTEGER *)&conv)->value;
>       }
>
>       break;
>
>     case FIELD_TYPE_INT24:
>     case FIELD_TYPE_SHORT:
>     case FIELD_TYPE_LONG:
>     /*case FIELD_TYPE_TINY:*/
>     case FIELD_TYPE_LONGLONG:
>     case FIELD_TYPE_YEAR:
>
>       GB.NumberFromString(GB_NB_READ_INTEGER, data, strlen(data),
> &conv);
>
>       val->type = GB_T_INTEGER;
>       val->_integer.value = ((GB_INTEGER *)&conv)->value;
>
>       break;
>
>     case FIELD_TYPE_LONGLONG:
>
>       GB.NumberFromString(GB_NB_READ_LONG, data, strlen(data), &conv);
>
>       val->type = GB_T_LONG;
>       val->_long.value = ((GB_LONG *)&conv)->value;
>
>       break;
>
>     case FIELD_TYPE_FLOAT:
>     case FIELD_TYPE_DOUBLE:
>     case FIELD_TYPE_DECIMAL:
>
>       GB.NumberFromString(GB_NB_READ_FLOAT, data, strlen(data), &conv);
>
>       val->_float.type = GB_T_FLOAT;
>       val->_float.value = ((GB_FLOAT *)&conv)->value;
>
>       break;
>

A bug! I forgot to remove the first one of course...

>   I think that I need also to add the new data type in the table_create
> function do you agree ?

Of course, you are right. I forgot to do that, whereas it is absolutely 
logical (I mean, if I change a function that converts SQL datatype to gambas 
datatype, I have to do the contrary somewhere else!)

>
>  I hope to have enought time to starting the test on the new datatype
> tomorrow. Have a nice day, Andrea.
>  Coai.
>

Regards,

-- 
Benoit Minisini