fastddsspy with idl type

[wannastar]

IDL include a structure HelloWorld, HelloWorld includes a struct point, point includes a struct timestamp(such as HelloWorld.idl),in this way, fastddsspy will come to Segmentation fault. However, if IDL include a structure HelloWorld, HelloWorld includes a struct point, point don't include any struc (like HelloWorld_OK.idl), fastddsspy will run ok.

HelloWorld_OK.idl
struct point{
short x;};
struct HelloWorld{
point pos;
unsigned long index;};

HelloWorld.idl
struct timestamp{
short h;};
struct point{
timestamp ts;
short x;};
struct HelloWorld{
point pos;
unsigned long index;};

step1:
fastddsgen -typeobjcet HelloWorld.idl

step2:
DomainParticipantQos pqos;
pqos.wire_protocol().builtin.typelookup_config.use_server = true;

eprosima::fastdds::dds::DomainParticipant* participant_;
participant_ = DomainParticipantFactory::get_instance()->create_participant(0, pqos);

eprosima::fastdds::dds::TypeSupport type_;
type_(new HelloWorldPubSubType())
type_.register_type(participant_);
type_->auto_fill_type_object(false);
type_->auto_fill_type_information(true);

step3:
HelloWorld hello_;
hello_.index(0);
hello_.pos().ts().h(10);
hello_.pos().x(10);
writer_->write(&hello_);

[wannastar]

it's same with this .
publisher: TypeLookupService -x complete
fastddsspy will come to Segmentation fault again.

[juanlofer-eprosima]

Hi @wannastar

What you are describing is one of the known issues with Dynamic Types currently present in master branch.
An exhaustive refactor tackling all these issues and aiming to fully comply with the OMG standard is currently being performed, and it is expected to be completed in the upcoming months.

For the time being, there exists an unsupported branch containing a bunch of temporary workarounds which probably would solve your issue.

[wannastar]

okay, i got it. i'm looking forward to a better fastdds tool.