CVS Head as of right now [after I fixed my dropped-lead-digit bug in the
apr_strtoi64()]...
Trying to serve a pdf, I added \"%{Range}i\" to my access log format just to see what
the
browser is requesting... the Range header request is for;
bytes=4096-
bytes=287090-288113, 279922-287089, 65536-91023, 91024-121023, 121024-151023,
151024-181023, 181024-211023, 211024-241023,
241024-271023, 271024-279921
Ok, the second is out of order, a general mess, but that's what Acrobat will do under
IE.
It's just fine by the rfc ... so first I tested and compared the first request for
bytes 4096-
and that first request worked just fine.
The second bytes= request sends back positively corrupted headers.
The complete server response (including what -should- have been headers) is over in
http://www.apache.org/~wrowe/byterange.results if you are interested in what happened
on win32. I want to check that this is specific to Win32, or have we gone and borked
all of our byterange handling somewhere along the way?
FWIW, the file being served is 288114 bytes. Any 288114 byte file should serve as a
study
of the bytes= requests above.
Bill
