Reading past max_samples_per_read

Hello Dallas,

The way to iterate over all instances, even if they have already been read, it is to use the operation read_next_instance() on the DataReader.  

Each time you want to begin an iteration you begin by calling read_next_instance() passing DDS_HANDLE_NIL as the value of the previous_handle parameter. This will give you all the samples associated with the first instance, then you call  read_next_instance() passing the instance_handle you get inside the DDS_SampleInfo as the value of previous_handle  and so forth until read_next_instance() returns DDS_RETCODE_NO_DATA.

Note however that you will be accessing samples ordered by instance, not ordered by the time when they were received. Samples returned on each call to   read_next_instance() will all be for the same instance and the will be ordered by reception time stamp. But samples returned by different calls to  read_next_instance() are not ordered.  

Gerardo

Hi Dallas,

I have some questions: do you want to read only non-read samples on every period? why would you not want to use the DDS_SampleStateMask when reading? I don't see that for the fact of reading periodically you couldn't use such filter. I think that would solve the problem.

Regarding the max_samples_per_read, you can also achieve same behavior by using the read's argument max_samples; losing the ability of configuring at run time though. In any case, an example of how to read up to max_samples_per_read is shown below:

MAX_SAMPLES_PER_READ = 1024;
...
/* execute every period */
...
FooDataReader_read(
    fooSeq,
    infoSeq,
    MAX_SAMPLES_PER_READ,
    DDS_NOT_READ_SAMPLE_STATE,
    DDS_ANY_VIEW_STATE,
    DDS_ANY_INSTANCE_STATE);
...

This would deliver the samples ordered by reception timestamp. If you want to get the samples ordered by instance, then you can proceed as Gerardo suggests and apply the sample state filter as well.

Antonio