Kenneth C. Arnold
kcarnold at arnoldnet.net
Thu Dec 28 10:48:14 PST 2000
According to Brian Havard (sometime around Thu, Dec 28, 2000 at 05:21:29PM +1000):
> On Thu, 28 Dec 2000 16:01:28 +1100, Michael Smith wrote:
> >>That should produce exactly the same thing as
> >>echo -e "GET /file.ogg HTTP/1.0\r\nAccept: */*\r\nUser-Agent:
> >ogg123\r\nHost: localhost(orwhatever)" | nc localhost(orwhatever) 80
> >>with the headers (everything before the first \r\n\r\n) stripped.
> >Indeed it should. Apparently, though, it isn't doing so. Hence my request
> >for more information (remember, this is on OS/2, not unix).
> I've tracked down the problem now. For some reason the 'seekable' test in
> ov_open_callbacks() (the seek_func(f,0,SEEK_CUR)) screws up the stream. If
> I take this out and force it to non-seekable it works just fine.
> Another small item, I had to add "fflush(opt.instream);" in ogg123.c at
> line 319 or the request didn't get sent and the 'Dump headers' code ended
> up in an infinite loop as it doesn't check for EOF.
Ooooops ... sorry :( Thanks for the reminder about HTTP being
nonseekable... actually when I get buffering working, I should allow a
seek within the buffer. Actually, that would make coding things a bit
more difficult (buffer is shared between two processes, and the writer
is in a tight write loop, so that would introduce more complexity), so
what do you say: is an in-buffer seek worthwhile?
Buffering is in an almost-working state here on my local copy ... but
I abandoned it in favor of partitioning a new drive and now homework
(argh!). I'll gladly get back to work as soon as possible; it's almost
there. btw, buffer on OS/2? How to do?
Kenneth Arnold <ken at arnoldnet.net> / kcarnold / Linux user #180115
<LI>application/pgp-signature attachment: stored
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 233 bytes
Desc: not available
Url : http://lists.xiph.org/pipermail/vorbis-dev/attachments/20001228/ce6f5bcf/part-0001.obj
More information about the Vorbis-dev