A bad merge, probably... This will show clearer the problem when there are not enough queues to serve a TX DPDK device. In this case we must rely on locking, but I don't want to start locking automatically. The user may augment the number of available queues with virtual devices.

Should fix #71

Fix isse with --enable-dpdk-pool because of a dependency loop

Signed/unsigned comparison, initialized after, ...

Make CLANG happy too

Make CLANG happy

- General documentation fixes
- Avalaibility of JSON library is exported
- HashSwitch is batch-compatible
- PaintSwitch is batch-compatible
- String have trim_space_left and replace
- Add statistics handlers for DPDK elements

Changes are mostly from Georgios Katsikas, RISE SICS and Tom Barbette (ULiege)

Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>

Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>

Do not initialize DPDK pool if we don't use --dpdk-packet-pool. This would not be bad IRL because most people compiling with DPDK actually use DPDK, but on travis it fails

Fix #78

Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>

Packet::make needs DPDK in that case

Solves #46

Imported from metron branch

Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>

This patch moves some code from the configure() function of
TimestampDiff to the initialize() function. This prevents
undesired access to queuedevice elements' (e.g., FromDPDKDevice)
internal memory, before such elements are initialized
Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>