suppressed so the never take up space/bandwidth in the VSL.

Reimplement vcl_trace using this:

	param.set vsl_mask +VCL_trace
	param.set vsl_mask -VCL_trace

XXX: doc-update needed

Some more text describing the behaviour on the various storage backend
+ some reformatting for readability

to like it...

Patch by:	daghf

assert before the parent carried out the assignment of the thread
id to the variable.

Detected by:	OSX

Spotted by: OSX pthread implementation

And yes, why ptrdiff_t even exists and why it can be different,
semantically or otherwise, from ssize_t is beyond me...

from sphinx. It was never properly maintained.

for ESI processing.

Polish the VGZ code a bit while here anyway.

Fixes	#1184

the xid, we already have that.

If people only specify Transient, run only on Transient.

Fixes	#1176

storage.

Fixes	#1176

are on the busy-obj waitinglists

(busyobj).

This replaces the current "VID" as visible in the X-Varnish header.

The new "stolen numbers" causes a lot of testcases to need updates,
most trivial, but a few needed more extensive work to be predictable
under the new circumstances.

Next step is to carry this change through in VSL, and link all
the VXID's to each other.

Submitted by:	Federico G. Schwindt

Submitted by:	Federico G. Schwindt

on kernels without coherent VM/buf (OpenBSD, still ?, really ?)

Submitted by:	Federico G. Schwindt

Old indentation made the :ref: into a literal block, thus not a real
reference. This should do the trick.

original ESI processed request, rather than to the included
requests.

Fix by allocating/cloning a new request for the esi:include
transactions.

Testcase by:	scoof
Fixes	#1168