why citus does not support OCLASS_OPCLASS in supportedDependecnyByCitus function

[jank9201]

recently, I use the bingo extension from https://github.com/epam/Indigo, which has the following definintion

CREATE OPERATOR CLASS bmolecule
FOR TYPE bytea USING bingo_idx
AS
        OPERATOR        1       public.@ (bytea, sub),
        OPERATOR        2       public.@ (bytea, exact),
        OPERATOR        3       public.@ (bytea, smarts),
        OPERATOR        4       public.@ (bytea, gross),
        OPERATOR        5       public.< (bytea, mass),
        OPERATOR        6       public.> (bytea, mass),
        OPERATOR        7       public.@ (bytea, sim),
        FUNCTION	1	matchSub(bytea, sub),
        FUNCTION	2	matchExact(bytea, exact),
        FUNCTION	3	matchSmarts(bytea, smarts),
        FUNCTION	4	matchGross(bytea, gross),
        FUNCTION	5	_match_mass_less(bytea, mass),
        FUNCTION	6	_match_mass_great(bytea, mass),
        FUNCTION	7	matchSim(bytea, sim);

, the above-mentioned defininition is not supported in source file src/backend/distributed/meta/dependency.c can someone tell me why such kind of class do not be supported?

[emelsimsek]

CREATE OPERATOR CLASS is not propagated as mentioned in #4812

[jank9201]

Hi, emelsimsek: Thank you for your reply, I've added a piece of code in the switch clause in supportedDependecnyByCitus function, so that OCLASS_OPCLASS case can return NULL; currently, it seems to be ok excepting for two warnings. if any other bug caused by such change found, I'll reported later, thanks again.

[jank9201]

Hi, emelsimsek: Thank you for your reply, I've added a piece of code in the switch clause in supportedDependecnyByCitus function, and OCLASS_OPCLASS case returned NULL; Currently, it seems to be ok excepting for two warnings. If any other bug caused by such change found, I'll reported later, thanks again.

[jank9201]

Hi, emelsimsek: Thank you for your reply, I've added a piece of code in the switch clause in supportedDependecnyByCitus function, and OCLASS_OPCLASS case returned NULL; Currently, it seems to be ok excepting for two warnings. If any other bug caused by such change found, I'll reported later, thanks again.