Fix error in read_sample()/take_sample() operations with Datasharing [12130]

[richiware]

The bug was reproduced when reader's history is 1 depth and while read_sample()/take_sample() operation is deserializing the data, the data is overwritten by the datawriter. The ReadTakeCommand returned OK and the code tries to access a 0 length vector.

Also the code was changed to return NO_DATA if the operation will return no data.

[MiguelCompany]

Also the code was changed to return NO_DATA if the operation will return no data.

Instead of doing this, we could add the following lines on DataReaderImpl::read_or_take_next_sample, right after ReturnCode_t code = cmd.return_value();

    if ((ReturnCode_t::RETCODE_OK == code) && (0 == sample_infos.length()))
    {
        code = ReturnCode_t::RETCODE_NO_DATA;
    }

[MiguelCompany]

@richiware I would also add a regression test for read_next_sample

[richiware]

@MiguelCompany Regression test was implemented.

About your conditional statement suggestion in the read_or_take_next_sample(), I'm not sure. It will only fix the error in the read_next_sample/take_next_samplefunctions.
In case of calling read/take function with one sample in the history and the deserialization of that change fails, the function will return OK.

[MiguelCompany]

LGTM with green CI

[MiguelCompany]

@richiware This needs a rebase

[richiware]

Rebase was done.

[MiguelCompany]

LGTM with green CI