from RFC5737's TEST-NET-1 and put a comment there to explain.

I fear this is a loosing battle, but at least this is more "official"
than the previous "10.255.255.255" choice.

Fixes #1159

Inspired by:	patch from geoff

size of the request, probably sleep copy&paste ?

Fixed	#1175
Submitted by:	martin

Increase the id field in VSM to 128 bytes to make space for 64 backend
VCL name + IPv4, IPv6 and portnumber.

This is the same as a9e5d350.

Fixes: #1224

Fixes #1167

With fix for #1275 we can now do a proper 503 when malformed
Vary-headers are received from backend. Error out instead of accepting
bogus ':'.

Conflicts:
	bin/varnishd/cache_vary.c

(out limit) characters long.

Fixes: #1274

Test case by: Dag Haavi Finstad

Fixes: #1275

Test case by: Dag Haavi Finstad

headers, and do 503 on parse error.

Conflicts:
	bin/varnishd/cache.h
	bin/varnishd/cache/cache_req_fsm.c
	bin/varnishd/cache_vary.c

This should not have any impact on the "normal" deliveries, as these
will as before be done through a single writev operation.

For deliveries involving chunked encoding (e.g. ESI or gunzip), this
should prevent an unnecessary delay where the Nagle algorithm kicks in
on the end-chunk write. This should give higher throughput and better
utilization of a single connection, especially for clients not doing
pipelining.

Conflicts:

	bin/varnishd/cache_acceptor.c

Fixes: #1261

This makes varnishncsa behave the same way as varnishlog, and prevents
overly large memory allocations if it should see a bogus fd (shmlog
overrun?).

XXX: This will not fix varnishncsa for current trunk, as that is using
XIDs instead of FDs breaking some functionality in the tools. This
will be handled in the API rewrite.

Fixes #1233

When looking at varnishlog it looks like it is send_timeout that was reached,
but in fact it is idle_send_timeout. Clarify the language so the user
understands which parameter to adjust.

Use VSL_Dispatch instead of VSL_NextSLT in varnishtop, which does this
automatically for us.  Clean up accumulate() a bit in the process.

Fixes #1213

This reverts commit 2587b1b9.

3.0 branch does not build with this commit.

Fixes #1194

for ESI processing. Adapted from phk's fix for #1184.

wrk_herder_thread and wrk_herdtimer_thread causing bogus n_wrk_max
stat increments.

Fixes: #1183

If the ban lurker finds a busy object, skip the object and defer the
checking to a later pass.

No test case because it is very timing sensitive (but see the trac bug
report for a VTC-file that mostly triggers it on the 3.0 branch only)

Fixes: #1162

There is a short window of opportunity between HSH_Insert() and
HSH_Unbusy(), where the locks are released and the expiry thread can
try to expire an object that is still busy (for objects with very
short TTL). This patch makes the expiry thread check for busy objects
and wait if the head of the binheap is busy.

No test case as the window is too small.

Fixes: #1150