You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The setting SET DECFLOAT BIND and isc_dpb_decfloat_bind only influence DECFLOAT types. It doesn't influence NUMERIC and DECIMAL with precision larger than 18. It would be helpful if the same support was extended to those types.
I see two options:
1. Extended coverage of 'decfloat bind' to include the extended precision numerical types
2. Define a separate bind setting (SET statement + dpb) for the extended precision numerical types (eg 'decfixed bind' or 'extended_numeric bind' or something like that)
The first (extend coverage of 'decfloat bind') has my preference as it is simpler, and drivers or end-users that need this will likely either support both types or none (as supporting DECFLOAT will automatically lead to support of the extended precision numerics and vice versa). On the other hand having a separate setting can be used for fine-grained tuning (eg using double precision for decloat, and char for extended precision numerical types).
I believe the issue is resolved after changing of an underlying type for high precision numeric to 128-bit integer and adding generic SET BIND statement
Submitted by: @mrotteveel
The setting SET DECFLOAT BIND and isc_dpb_decfloat_bind only influence DECFLOAT types. It doesn't influence NUMERIC and DECIMAL with precision larger than 18. It would be helpful if the same support was extended to those types.
I see two options:
1. Extended coverage of 'decfloat bind' to include the extended precision numerical types
2. Define a separate bind setting (SET statement + dpb) for the extended precision numerical types (eg 'decfixed bind' or 'extended_numeric bind' or something like that)
The first (extend coverage of 'decfloat bind') has my preference as it is simpler, and drivers or end-users that need this will likely either support both types or none (as supporting DECFLOAT will automatically lead to support of the extended precision numerics and vice versa). On the other hand having a separate setting can be used for fine-grained tuning (eg using double precision for decloat, and char for extended precision numerical types).
Commits: FirebirdSQL/fbt-repository@f87a8b6
====== Test Details ======
See fbt-repo/tests/functional/datatypes/decfloat-binding-to-legacy.fbt ( statement: "set bind of numeric(38) to numeric(18)" )
The text was updated successfully, but these errors were encountered: