Add typename to reduce compilation warnings

[galou]

The warning appears on Humble at least.

[peci1]

Thanks for the PR.

Please, fix the suggested code style violation so that we can see test results. But the tests in this repo are actually only linters, so I guess that the code compiles fine on all platforms. But I'd like to see the green marks ;)

I have one comment though - we don't support Humble at all. We have no robot running it. All our codes are Jazzy+ . If this PR doesn't seem to break anything and it works on our robots, I'll merge it, but please don't count on Humble support - we might break it any time.

Are you actually using this driver, or just experimenting with it? So far we haven't avoided behavior changes and such things in the driver, because we thought it's only our group using it. But if more people find it useful, we can certainly start behaving in this repo =)

[galou]

Hi Martin, I didn't want to use Humble but switching our robot to Jazzy would have taken me too long (I wanted this and other things working within Monday). I misbehaved even more than you because I delete from our forks (this and the dependencies) the things that didn't compile. I was actually surprised how many changes are necessary.

I'm not sure whether I need your driver or not (I would rename it filter by the way) but I saw that it has one thing I needed, outputting a sensor_msgs/Imu for robot_localization. We have a dual-antenna setup and I didn't find any other solution. In the meantime, I saw that the Septentrio driver is able to output a geometry_msgs/PoseWithCovariance message and implemented a processPose function because their Pose is not a Pose but just a container for WGS84 coordinates! I actually realized that there's no pose publisher. By trying to solve the issue I tried to open the configuration page but couldn't load the web interface. To make the story short, we probably have a power issue (or maybe because we started the units without GNSS signal) and the units booted sometimes in "Firmware ugrade" mode (Septentrio is prepared for this). Unfortunately, I upgraded to 4.15 which introduced authentication on TCP ports and our setup doesn't work anymore, rover doesn't get the correction from mobile base ( I wrote an issue on the customer portal). I'll solve this and Jazzy next week.

Does the issue with DO-NOT-USE values persist on recent firmware? It'd be a good idea to document which versions of the firmware and ROS driver are concerned with this issue.

[peci1]

outputting a sensor_msgs/Imu for robot_localization. We have a dual-antenna setup and I didn't find any other solution.

We also use it for this purpose. Just beware that if the initial orientation is bad, r_l (or navsat_transform_node) has a hard time correcting that. That's why we have so strict thresholds on variance by default.

Also check out the newly added heading filter. We've seen cases when the heading swapped 180° for a few measurements. Not that r_l would be happy from these swaps...

their Pose is not a Pose but just a container for WGS84 coordinates! I actually realized that there's no pose publisher

This is weird. They usually duplicate all outputs, once in spherical and once in cartesian (UTM) coordinates. Did you check the cartesian outputs?

the units booted sometimes in "Firmware ugrade" mode

I've seen this quite rarely with our units. A power problem is likely.

Unfortunately, I upgraded to 4.15 which introduced authentication on TCP ports and our setup doesn't work anymore, rover doesn't get the correction from mobile base ( I wrote an issue on the customer portal)

I'm glad I feared this update so far :-D Could you be more specific on which TCP ports need auth? Like NTRIP?

Does the issue with DO-NOT-USE values persist on recent firmware? It'd be a good idea to document which versions of the firmware and ROS driver are concerned with this issue.

I think in 1.4.0 most of the DNU values were changed to NaNs. I think I still saw some after the update, but quite rarely. So some part of the filter is probably not required anymore, but we kept them because it's easier for us.