Why do we use Rviz when we have Gazebo ?

edit

So why do we need a visualization tools when we already have a real world simulator with visualization tool.

because most of the times we have the actual real world (in which our robots walk/fly/drive/crawl/swim/move/etc), so we don't need a simulation of it, but we would still like to visualise data going around in our ROS applications.

Edit: some clarification: the point here is that Gazebo couples the visualisation with the simulation, which makes it near impossible to use the visualisation component without the simulation (yes, gzserver and gzclient are separate binaries, but they're still rather tightly coupled). So with that in mind, it makes sense to have a separate visualisation tool (RViz in this case) that can be used without the simulator. Note also that RViz and Gazebo have always been separate projects, they just happen to both have been developed/curated/maintained by WillowGarage/OSRF/OR at some point.

Most use-cases for ROS actually wouldn't even include a simulator (any simulator, not just "not Gazebo"), so forcing people to use that would not make much sense.

If with your question you implied that it would be good -- from a software re-use perspective -- to not duplicate the visualisation aspects of RViz and the Gazebo UI, then I would fully agree with you. Unfortunately it would seem the ROS and Gazebo development teams don't share that opinion (or don't seem to want to act on it).

Edit (2021-08-17): just remembered that as part of GSoC 2020 a re-implementation of RViz using Ignition (~= Gazebo) technology was created: GSoC 2020: Ignition RViz. While it hasn't replaced RViz, it does show that it would be possible to reuse many parts and avoid the (perceived) duplication of effort.

Repository: ignitionrobotics/ign-rviz.