Progress callback called with UL 0 out of 0[LF]
-Progress callback called with UL 8 out of 0[LF]
-Progress callback called with UL 16 out of 0[LF]
-Progress callback called with UL 26 out of 0[LF]
-Progress callback called with UL 61 out of 0[LF]
-Progress callback called with UL 66 out of 0[LF]
+Progress callback called with UL 40 out of 0[LF]

I more or less understand the differences here now. The test has this data that gets sent:

static const char * const post[]={
  "one",
  "two",
  "three",
  "and a final longer crap: four",
  NULL
};

Vanilla curl gets an update for each entry, which is why there are five entries after the initial UL 0. The length of each update is the length of the string, plus another 5 or 6 chars for a hexlen and some EOL chars. That's how we end up with (3+5) + (3+5) + (5+5) + (29+6) + (0+5) = 66 bytes.

curl+hyper ends up doing things slightly differently: it doesn't include the hexlen/EOL chars, so we end up with 3 + 3 + 5 + 29 + 0 = 40 chars. Also, it doesn't call Curl_pgrsUpdate between each Curl_pgrsSetUploadCounter, so we get a single update at the end instead of multiple updates.

I don't know if these differences are important. Maybe it's worth having different expected output for the hyper build?

Thanks!