tests using stream t2 vs tests using chunked content

[jpiesing]

Stream t2 is the version of the chunked content before the chunking script is run on it.
4 generated tests include t2 in the filename.

• buffer-underrun-and-recovery__t2
• low-latency-short-buffer-playback__t2
• ocal-low-latency-initialization__t2
• local-random-access-from-one-place-in-a-stream-to-a-different-place-in-the-same-stream__t2

2 generated tests have chunked in their name

regular-playback-of-chunked-content__chunked
regular-playback-of-chunked-content-non-aligned-append__chunked

t2 is not meant to be used without being chunked. Is is correct that the 4 tests are using different content from the 2?

• If yes then this needs to be revisted and the 4 should be changed to using chunked
• If no then the filenaming is really confusing

Also note that TVs number 2 and 4 from the London plugfest decode all frames from the __chunked tests but drop a lot of frames at the start of the __t2 tests.

[gitwjr]

From the DPCTF Call on 2024-04-10
Fraunhofer regenerated using content t1. No difference in P/F or the numbers were found using t1 vs t2. Needs deeper analysis to determine why it is failing. Noted that this test has far fewer failures on a Desktop than on TVs. They will rerun and review the results after changes from issues # 123, 122, 121 and compare the new results with results from testing with a Desktop browser.

[jpiesing]

2024:05-14: On one of the recordings from the London plugfest, this still fails due to a small error in the duration observation. We should wait for Fraunhofer to run the tests on their small zoo and see how common this error is.