On 01/07/2019 08:26, Jiří Činčura wrote: >> I would report back when I do the test, as time permits. > OK, I have some numbers. Take it as it is. The absolute value is probably not > interesting, but the ratios are (with PSQL as a baseline). > > | Method | Mean | Error | StdDev | Ratio | RatioSD | > |------- |----------:|---------:|----------:|------:|--------:| > | PSQL | 265.30 ms | 3.572 ms | 3.3409 ms | 1.00 | 0.00 | > | UDR | 311.96 ms | 2.880 ms | 2.6942 ms | 1.18 | 0.02 | > | UDF | 93.83 ms | 1.205 ms | 0.9407 ms | 0.35 | 0.01 | > > This is just the invocation. Nothing meaningful was in the body. > The current implementation of UDR has a "PSQL" stub.
It's used to validate parameters and return values in relation to constraints, for example. The stub was used due to inflexibility to directly call the internal functions. Maybe this should be reviewed and optimized. Adriano Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
