On 12/19/2016 12:33 AM, Jeon wrote: > As I understand, at first run (n_ofdm_symbols = 0), it reads `n = > symbols_to_allocate` tags from input and puts it into `k = > nitems_written(0)`, which is the starting position of sync word. At > second run (n_ofdm_symbols = 1), it reads tags again, and puts it into > `k = nitems_written(0) + 1 + d_sync_words.size()`. Hence, the first > group and the second group of tags are apart by offset `d_sync_words.size()` > > Is it my understanding of source code correct? If so, I wonder whether > there is a reason for placing the first group of tags onto sync word, > not onto the first data ofdm symbol. If I miss something in interpreting > the code, please let me know it.
There is a reason to do this, but there's also an issue. The idea is to put tags as close as possible to the samples they were on the input stream, *except* if they came in on the first sample, because in that case, we make the assumption that they were, in fact, meant to be at the front of this tagged stream. The issue is that it'll do this for any tag that's on the first OFDM symbol. The correct test would thus not be n_ofdm_symbols == 0, but rather tags[t].offset - nitems_read(0) == 0 or whatever correctly tests for if the tag is actually on the first sample, not just any sample in the first OFDM symbol. Cheers, Martin _______________________________________________ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
