Correctly inform that a BoundedSequence is bounded #71
Conversation
Signed-off-by: Miguel Company <MiguelCompany@eprosima.com>
| @[ else]@ | ||
| size_t array_size = 0; | ||
| full_bounded = false; |
There was a problem hiding this comment.
@MiguelCompany because it is unbounded, right?
| @@ -530,12 +530,11 @@ size_t max_serialized_size_@('__'.join([package_name] + list(interface_path.pare | |||
| size_t array_size = @(member.type.size); | |||
| @[ elif isinstance(member.type, BoundedSequence)]@ | |||
| size_t array_size = @(member.type.maximum_size); | |||
| is_plain = false; | |||
There was a problem hiding this comment.
@MiguelCompany because it is too early to know, right?
There was a problem hiding this comment.
Because any sequence will mean the type is not POD, and this is already covered below on the if isinstance(member.type, AbstractSequence)
| @[ end if]@ | ||
| @[ if isinstance(member.type, AbstractSequence)]@ | ||
| full_bounded = false; |
There was a problem hiding this comment.
@MiguelCompany because it may be abstract and bounded, right?
There was a problem hiding this comment.
Because BoundedSequence inherits from AbstractSequence and should return full_bounded = true
|
CI up to
|
|
CI again:
|
|
CI again as it's been a while:
|
This PR fixes a mistake I made on #67, that made bounded sequences be treated as unbounded ones.
The full fix also needs ros2/rmw_fastrtps#540, so that serialization sizes are calculated correctly.