We can already deduce the name expected by DPKG or RPM from what we
have.

With more comments this time!

As per discussion, to better facilitate alternate init-
systems and re-packaging efforts, we want to keep in the
sysv versions of the packaging files.

In other words, we get that part of the release tag from upstream
already.

Same as traditional shared objects, it now looks like this:

    varnish = x.y.z-r
    varnish(x86-64) = x.y.z-r
    libvarnishapi.so.1()(64bit)
    libvarnishapi.so.1(LIBVARNISHAPI_2.0)(64bit)
    varnishd(abi) = sha1
    varnishd(vrt) = a.b
    vmod(blob)(64bit) = x.y.z-r
    vmod(directors)(64bit) = x.y.z-r
    vmod(purge)(64bit) = x.y.z-r
    vmod(std)(64bit) = x.y.z-r
    vmod(vtc)(64bit) = x.y.z-r

Maybe varnishd(*) provides should be ISA-bits-aware too.

Since we wrap calls to the original find-provides it loses the ability
to do that so it becomes our responsibility.

A bit fragile when it comes to grabbing the version.

This change aligns DPKG provides to their RPM counterparts with one
notable exception: the ABI being a hash, it may start with a letter
and go against DPKG's policy. Since it's only used for strict equality
this is not a problem.

The new provides look like this:

    Provides:
     libvarnishapi1,
     varnish,
     varnishd-abi-sha1
     varnishd-vrt (= x.y)

Partially fixes #8 because in order to make this useful for VMODs
without forcing package maintainers to manually require either the
exact ABI or a given VRT version, a means for automatic requires
for VMOD builds is missing. This doesn't address the NOGIT case,
but we don't support packaging not originating from a git commit.

Unlike RPM packaging, this does not list installed VMODs, as I'm
not certain it can be done in the first place (probably though).

Refs c6d66785
Closes #92

Maybe we should rename common/ to systemd/ as well.

While at it, add some polish to the overall script and simplify the
different cases.

Still not targetting el4.

No compatibility is needed for 4.0, and removing the RST sources creates
dead ends for all the "Show Source" links present on every pages built
by sphinx.

The find-provides script takes the original script and forwards almost
all lines to the original find-provides command provided by RPM. In
order for this to work, %global needs to be used instead of %define to
ensure immediate expansion and avoid infinite recursion.

VMODs are not forwared to find-provides because it matters little to
advertise them as shared objects in the traditional sense.

The output looks like this:

    vmod(blob)
    vmod(directors)
    vmod(purge)
    vmod(std)
    vmod(vtc)
    varnishd(abi) = <sha1>
    varnishd(vrt) = x.y

Partially fixes #8 because in order to make this useful for VMODs
without forcing package maintainers to manually require either the
exact ABI or a given VRT version, a means for automatic requires
for VMOD builds is missing. This doesn't address the NOGIT case,
but we don't support packaging not originating from a git commit.

This would make #92 obsolete once something similar is done on the
DPKG side.

We have auto-requires for DSOs we link to.

Otherwise it doesn't match for weekly packages.

It wasn't meant to be checked in, but it yields slightly faster mock
builds on my system. And for some reason mock doesn't always kill
systemd-nspawn containers and I don't have time to debug that now.