How to select the best PUT?

[git-afsantos]

I think it makes sense for the PUT to be whatever is in fix.fix-in, in the bug description, as it is where changes were made.

I am opening this issue just to gather an opinion on whether it is relevant for the bug.package packages to be considered the PUT.

Take Kobuki e3c9bbc.bug as an example. The issue is that there was no source link in the rosdistro YAML. I assigned the missing Kobuki packages under bug.package, but the fix happens in rosdistro.

I can either use Kobuki's packages as PUT, and the test script gathers the rosdistro YAML somehow, or I can use rosdistro as the PUT, making sure the files are there and can easily be read/tested.

[gavanderhoorn]

Personally, I feel e3c9bbc should not have been included in the bug set.

It's not a bug. It's a missing bit of yaml in the rosdistro database.

---

And it seems "we" agreed with me already:

We decided to not reproduce this bug, as this is not really an interesting case from the perspective of software analysis. Besides, the fix occurs in the rosdistro repository, which does not contain a ROS package, and makes the reproduction process harder.

[gavanderhoorn]

So returning to the question:

How to select the best PUT?

I would say that the PUTs (plural, as that is now supported) should be the packages in which changes to the sources (in the broadest sense, so .xml, .cpp, .h, .hpp, .urdf, etc, etc) are made to fix whatever the bug was.

#162 tracks the cases where PUTs from multiple repositories are involved in a single bug, and I believe we have (almost) all the machinery to reproduce those as well.

[git-afsantos]

Should we remove e3c9bbc so as not to cause confusion, and since it is clearly an outlier? Or do we leave it precisely as an example of what not to reproduce?

[gavanderhoorn]

I'm not sure. Leaving it makes as much sense to me as removing it.

I was not the one who added e3c9bbc, so perhaps @wasowski or @brabrand can clarify what the intention was?

[git-afsantos]

Actually, I was the one who added it. I believe at the time we were not yet discussing bug reproduction, otherwise I would have skipped it. I think the goal at the time was to collect bugs as diverse as possible, but it is debatable whether this can be considered a bug.

Edit: I am in favour of removing it, since it does not really seem to fit into what we are doing here.