Hello Michaël,
>> I'll be happy if you do the review of the resulting split.
>
> OK, I am fine with this scenario as well. I have luckily done nothing yet.
Here is a 3 part v29:
a) add support for integer functions in pgbench, including turning operators as functions, as well as some minimal
infrastructurefor additional types (this allows to minimize the diffs with the next patch which adds double).
b) add support for doubles, including setting double variables. Note that variable are not explicitely typed. Add
some double-related functions, most interesting of them for pgbench are the randoms.
c) remove \setrandom support (as thanks to part b \set x random(...) does the same). Prints an error pointing to the
replacementif \setrandom is used in a pgbench script.
--
Fabien.