Move to a new recorded data table about pyscada HOT 4 OPEN

a float64 floatfield has only 52 bits for the number the rest is for the exponent, so only 32 bit integers can be stored, this is Okay for most values, but we have to drop INT64 and UINT64 data types. I'm not sure what would be the performance penalty for having a char or textfield as column in the DB, but i suspect for bigger queries it would be not so good. With the new option to have more than one data table/backend would it be possible to have a separate Table for variables that have to store data as str?

from pyscada.

Ok, so we need 3 fields:

• BigIntegerField for integers (with the shift technique for unit64),
• FloatField for floats,
• CharField or TextField for strings.

I don't know if there is a performance penalty to have a char or text column for the variable not having data in it, I will try it.

One solution with the new data source architecture could be to separate this in 3 tables:

• RecordedDataInteger,
• RecordedDataFloat,
• RecordedDataString.

One question is what should the manager do if the variable value class is:

• string and the value to store is an integer or a float,
• float and the value to store is an integer or a string,
• integer and the value is a float or a string ?

from pyscada.

One question is what should the manager do if the variable value class is:

in my opinion either drop the value and throw an error or convert to the appropriate data type

from pyscada.

Ok so finally all the values of a variable will be in only one table.

from pyscada.