Include type-info in sertype_equal for default sertype implementation #1698
Conversation
When comparing sertypes that have xtypes type meta-data, the type information should be included. This commit adds the check to the default sertype implementation and adds a test for this. Signed-off-by: Dennis Potman <[email protected]>
| if (a->type.flagset & DDS_TOPIC_XTYPES_METADATA) | ||
| { | ||
| ddsi_typeinfo_t *ti_a, *ti_b; | ||
| ti_a = ddsi_typeinfo_deser (a->typeinfo_ser.data, a->typeinfo_ser.sz); |
There was a problem hiding this comment.
Do we really need to deserialize here? Or can we get away with comparing the serialised representation?
For that, we would have to be certain about the endianness and padding, and perhaps something else. The first two are reasonably easy to deal with, the third is of course tricky 😄
There was a problem hiding this comment.
The problem is the set of hash type identifiers the type depends on, that is included in a type-info object. The spec (7.6.3.2.1) does not specify any ordering for this list (TypeInformation.[minimal/complete].dependent_typeids), and it is also allowed to leave out one or more type identifiers, as long as the dependent type count is correct (TypeInformation.[minimal/complete].dependent_typeid_count).
When comparing sertypes that have xtypes type meta-data, the type information should be included. This PR adds the check to the default sertype implementation and adds a test for this.