- 26 Jan, 2011 1 commit
-
-
Per Buer authored
-
- 25 Jan, 2011 10 commits
-
-
Per Buer authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
all of it for gzip buffers.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Tollef Fog Heen authored
The tarballs live on repo.varnish-cache.org, not www.varnish-software.com any longer, so adjust spec file to match.
-
- 24 Jan, 2011 14 commits
-
-
Poul-Henning Kamp authored
my section of hell for the dimwits who decided that switching off_t from 32 to 64 bits was too scary to contemplate.
-
-
Kristian Lyngstol authored
-
Kristian Lyngstol authored
Essentially ignores everything git status would list after a normal build.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
vary-matching. Varnish will adapt whatever we find to the clients abilities as necessary.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
after vcl_recv{}. This is necessary to get consistent Vary: header processing.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
host system.
-
Poul-Henning Kamp authored
from a non-gzip ESI object or during gunzip delivery.
-
- 22 Jan, 2011 15 commits
-
-
Poul-Henning Kamp authored
Try again: Add the necessary code (& a testcase) to pad out an included gzip file, no matter which of the 8 bits of a byte the last bit lands in.
-
Poul-Henning Kamp authored
gzip file, no matter which of the 8 bits of a byte the last bit lands in.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
residual stopbit we expect.
-
Poul-Henning Kamp authored
to come.
-
Poul-Henning Kamp authored
indeed gzip'ed data and extracts the magic bits should it ever be included from ESI. In general Varnish is not in the business of inspecting content and with the exception of ESI we still do not do so. What we do here is make sure that we do not serve corrupt data, in the same sense as we bail out on chunked encoding trouble or HTTP header trouble from the backend.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
For a limited number of cases we can now esi:include a gzip'ed object into a gzip'ed ESI object. There is less than 1 chance in hell this will work for you in practice, I need to handle the N-1 other cases of last bit and stop bit combinations in the gziped object, but that is just a small (but tedious) matter of programming.
-
Poul-Henning Kamp authored
objects when we know them. For now: use malloc for gzip instead of running out of workspace.
-
Poul-Henning Kamp authored
in an gzip'ed ESI object.
-
Poul-Henning Kamp authored
ESI object.
-
Poul-Henning Kamp authored
in a gzip'ed ESI object.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-