This is in essence the same VMOD so migrating from one to the other
should result in no VCL change, except for the removal of DEPRECATED
cookie.filter_except().

The test suite was refreshed to take advantage of recent features and
mostly operates in vcl_synth as a result, not wasting cycles bringing
needless backends up.

The C code and VCC descriptor were also improved for better code style
compliance and documentation consistency, without changing the VMOD's
behavior. On the C side actual changes mostly consisted in adding missing
assertions but there were notable changes:

- usage of VRE_Free instead of free (with test case)
- WS_VSB_* facility usage for the one use case

There are other opportunities for improvement, like usage of more modern
facilities like VRT_fail, or a named Lck instead of a plain mutex.

Closes #3184

or rather: fail the task if being rolled back

I forgot the namespace when I first split the test suite.

We only need the @VTC_TESTS@ substitution, and doing so saves around
13kB per Makefile.

That part of the automake boilerplate is generated conditionally, and
since we check automake_boilerplate.am files in the list of tests is
printed with one per line to be diff-friendly.

A test was "randomly" chosen to be the first VMOD-specific test.

Generated automake_boilerplate.am files no longer start nor end with
empty lines.

This is probably going to mandate some changes in vtest to collect
VMOD-specific test suites.

Refs #3184

The effect is currently the same, but for VMODs we shouldn't need a
complete configure step.

Refs #3184

With a growing list of bundled VMODs we want to split the main test
suite to colocate VMOD-specific tests with the rest of its sources.
Because a VMOD test case may involve other binaries, and will always
involve at least varnishtest and varnishd, we insist that make check
kicks in only after a complete recursive build.

This doesn't work in this scenario:

    cd lib/libvmod_<name> && make check

The assumption is that at this point a full build was already performed
and changes are only happening in that sub directory.

Refs #3184

In maintainer mode, whenever a file managed by autoconf or automake is
changed it is regenerated on the fly. For Makefile.am files this also
applies to any fragment .am file referenced by the include directive.

The configure script disables the maintainer mode by default, which is
probably a sane behavior because downstream usage of dist archives don't
need the various autotools to be present, unless a Makefile or similar
is patched.

This is explicitly enabled by the autogen.des script, in order to make
this change only impact development, saving most of the time a configure
step after amending something in the build.

Refs #3184

RX_TIMEOUT was issued when the timeout_idle was hit waiting for another
client request and accounted as an error in sess_closed_err.

We now turn this case into an RX_CLOSE_IDLE which does not get accounted
as an error.

Closes #3208

I don't remember how I ended up with its private counterpart,
conditionally even, but this has been part of automake since
version 1.10 so there's no reason to beat around the bush.

... and ensure it always gets set during cli operations also

we used vcl_max both for the max vcl version and the maximum number of
vcls to load.

read the actual directories and filter on prefix strings.

Note XXX relative to #3202

Usage pattern:

     struct vsb vsb[1];
     char *p;

     WS_VSB_new(vsb, ctx->ws);
     VSB_printf(vsb, "blablabla");
     p = WS_VSB_finish(vsb);
     if (p == NULL)
             return (FAILURE);

Does this stabilize a vtest false negative which I do not understand?

There are two one-second delays on the server side, but varnishd might
read the response a bit longer after the first delay such that the
second interval is shortened from the varnish perspective.

Ref #3210

Other man pages seem to have bogus dependencies.

And harmonize the reStructuredText format of default values entries.

The assert on WS_ReserveSize() in ses_handle() can not trip because
sizeof (struct pool_task) is less than sizeof (struct waited). But to safe
guard against future problems if that were to change, this patch makes
sure that the session workspace can hold the largest of them before
entering the waiter, erroring out if not.

Proxy TLVs claiming to have PP2_TYPE_SSL sub-TLVs without complete payload
would cause a Varnish assert. This patch fixes the parsing of the TLVs.

h2_init_sess can only be reached through H1 with either previous knowledge
or opportunistic upgrade. Because of this the proto_priv session attribute
will always be set before entry. This patch simplifies and removes dead
code containing a call to SES_Reserve_proto_priv.

Note: Better diff with the --ignore-all-space option

In h2_init_sess, an extra call was always made to SES_Reseve_proto_priv(),
even though it was already reserved. This wasted a pointer worth of
session workspace. This patch removes the extra call.

If proxy protocol is in use, it is possible to fill the session workspace
exactly before entering http1_new_session(), which will cause it to assert
when calling SES_Reserve_proto_priv().

with this patch we will close the session gracefully.