hi everybody,
I want to
struct task {
long a;
long b;
sequence<double> cSeq;
};
struct entity {
long id;
long num;
sequence<task> myTaskSeq;
};
struct myType {
double speed;
sequence<entity> myEntity;
};
but in task struct,the cSeq's length is not sure,even for 0.
it will lead to takeup too much memory.
how can I improve this situation?
Hello,
The IDL you show above will produce erros when you run rtiddsgen on it. Sequences must be always templatized by the element-type. In addition it is possible to specify a maximum element count. If left unespecified the middleware defines a max element count of 100.
I have changed the IDL to the one shown below which can be parsed successfully by rtiddsgen.
struct task { long a; long b; sequence cSeq; }; struct entity { long id; long num; sequence myTaskSeq; }; struct myType { double speed; sequence myEntity; };I run rtiddsgen on the IDL above. Then I inserted the following code in the example application:
int maxSerialized = myTypePlugin_get_serialized_sample_max_size( NULL, RTI_FALSE, 0, 0); printf("maxSerialized = %d\n", maxSerialized);When I run this I get a maxSerialized = 4121212 bytes (4.12.MB) . The reason is that if the maximum length of the sequence is unspecified, then the miffleware assumes a size of 100 elements. This may not sound like much, but when you nest a sequnce inside a sequence inside a sequence you are getting a maxsize ~ 100x100x100x side of innermost element. In this case the innermost element is a long (size 4) so we get a number in the order of 4x100x100x100 ~= 4MB
This is just the size of a single elememt. Depending on how you have configured (in terms of resource limits) your DataWriter and DataReader you well get something like max_samples x myTypePlugin_get_serialized_sample_max_size(...). A max_samples of 1000 is not uncommon. If you did that you would get to 4GB of memory...
So this data-type is quote inneficient in terms of memory. Can you explain a little more of what you are trying to do? Khowing this perhaps we can suggest something more efficient way to define the data-types...
Regards,
Gerardo
hi Gerardo Pardo,
Sorry, I didn't make it clear.
I create a nested data structure which has many layers,like the above example. then,i have a mistake.It tell me is too larger,more than 3GB!
In fact,some of these sequences,don't need such a big length.
for exemple,the struct named task,may not have the property named cSeq.or the length of cSeq is very small!
The length is dynamic.
what can i do for it? thank you very much!
i want to change the length of the sequences dynamic,but i don't know how to do!
Or you can give me other advice, will solve my problem.I'm enormously grateful for your help!!!
Hi
when you create your IDL sequences without giving a count:
sequence<double> cSeq;
the default sequence length is 100. So you are telling the type system you need a sequence which will hold a maximum of 100 doubles. This default length is configurable at the command line.
If you give the sequence a value:
sequence<double, 5> cSeq;
You are telling the Type system that at most this cSeq sequence will need to hold 5 doubles.
When the cSeq is serialized, if it is empty, then NO SPACE is sent for that field. So this is good use of bandwidth.
See
http://community.rti.com/rti-doc/500/ndds.5.0.0/doc/pdf/RTI_CoreLibrariesAndUtilities_UsersManual.html#page_40
http://community.rti.com/rti-doc/500/ndds.5.0.0/doc/pdf/RTI_CoreLibrariesAndUtilities_UsersManual.html#page_74 (section 3.3.1.1)
hi, what you said about the sequences i know.
now i want to change the length dynamic.
because,the count of the cSeq may be 5,but may be 500!
If i give the sequence a value:500!
the struct
myType will too large more than 3G,becausethe nested structure!!!I felt mortified for this situation!
Hello Christine,
maybe this post can help: http://community.rti.com/comment/391#comment-391
Best,
Gianpiero