First, there may be multiple tags per sample. Use “get_tags_in_range” or “get_tags_in_window” to get all the tags you need. Using “get_tags_in_window(0, noutput_items, …)” will get you all the tags associated with samples passed as your inputs. (Perhaps you know this already and were just choosing a toy example, but I wanted to make it clear).
There is a pruning routine in the scheduler that will toss out “old” tags, i.e., tags that have been accounted for based on “nitems_read”, which is updated when you call consume_each. So I think you will end up getting the last 5 tags again on the next call to work. But you have already propagated them, so you will be adding duplicates. The solution is to only propagate tags in the range that you intend to tell the scheduler you’ve consumed. (Caveat emptor: I haven’t tested this out, but I believe this is how it would work). Sean From: [email protected] [mailto:[email protected]] On Behalf Of Richard Bell Sent: Monday, July 20, 2015 3:51 PM To: [email protected] Subject: [Discuss-gnuradio] Tag Propagation Question If I set_tag_propagation policy(TPP_DONT) and then re-add all tags in range to the output, but only produce a portion of the input items to output, what effect does this have? Here is a good corner case example of what I mean: Suppose I'm given 10 items, and every item has a tag associated with it. I then make the following actions set_tag_propagation policy(TPP_DONT) for( ii = 0; ii < 9; ii++) { add_item_tag(0, tags[ii].offset, tags[ii].key, tags[ii].value, tags[ii].srcid); } consume_each(5); return 5; I'm concerned this might cause tags to be added or lost. Thanks, Rich
_______________________________________________ Discuss-gnuradio mailing list [email protected] https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
