Create image for geometry2/12605ab

[wasowski]

I dared to commit this one to master, as I am reasonably certain of the description (even if not finished). This one is from Spring 2015, so we have some hope for the time machine working, despite the current limitations of our time travel technology :)

I think that I can reproduce it (well, I might need to produce a SIGINT against some core node in the test, but at least I know the direction what to think about unlike for many others :) )

[gavanderhoorn]

So the fix is from 2015-03-12T01:33:01Z, but the actual report is from 2013-09-17T13:34:49Z. Are we ok with a .rosinstall file for 2015?

[gavanderhoorn]

Btw: could I ask you to link to the bug description file when asking for a .rosinstall file? :) makes it easier for me to find the issue/pr link.

[wasowski]

Sure. I will try. The mode of work for tf2 is a bit different, so I am still learning. I think it makes sense to try to test them sooner than later, thus I make these issues. This is because, I still remember what they are about (unlike the UR bugs that are completely alien to me now, so they can just as well wait, or be taken by someone else...).

Re the date: yes this is my assumption that we should be able to reproduce it based on the time of the parent of the fixing commit, not on the reporting the issue time. I kind of expect that this was still a problem, when they were fixing this.

[gavanderhoorn]

@wasowski wrote:

Re the date: yes this is my assumption that we should be able to reproduce it based on the time of the parent of the fixing commit, not on the reporting the issue time.

Ok. Is that something for tf2_ros specifically, or something we should do in general?

Up till now I've always used the date of the issue that reported the problem. That seems more appropriate to me, as that would be the state of the system of the reporter at the time he encountered the issue.

[wasowski]

OTOH, using the fix time seemed more correct to me, from the selfish perspective that the fixing commit is then more likely to apply to our buggy snapshot :).

Honestly, I don't think this matters. We only promise to reproduce bugs. If we reproduce them, it should not matter on which precisely commit. It should just be recored (and it will be using tags or whatever data we have). So if it is easier to stick to the reporting time, I suggest to do so, and then only manually intervene in unusual cases, like this one.

[gavanderhoorn]

See e15feaae6d2f388426d6ae71cd06c56b4f1829c7.

[gavanderhoorn]

I've not added the bugzoo metadata. It's indigo though, so fairly straightforward.

[wasowski]

Thanks. I think I will just tag @ChrisTimperley

[wasowski]

btw. this was super fast. It took only 5 minutes?

[gavanderhoorn]

It looked like 5 mins to you, I have a time-machine :sunglasses: