Pull request: https://github.com/aibasel/downward/pull/10

Sorry my comment about the pull requests looking swapped is incorrect. Actually both pull requests (9 and 10) remove the task proxy and don't have other changes.

We had a look at this issue and pull request while triaging pull requests.

It looks like the pull requests for this issue (https://github.com/aibasel/downward/pull/10) and the one for issue906 (https://github.com/aibasel/downward/pull/9) were swapped, since pull request #9 is the one actually removing the TaskProxy.

To me, the open question is still what Malte said in msg8734: do we actually want to go in this direction? Not storing the TaskProxy together with the PDBs means that the caller has to know where the PDBs come from and make sure to not use the wrong task when accessing them.

Silvan, do you still want to push this? If so, can you add some arguments about why we should move in this direction? If this issue is just a prerequisite for issue906, maybe we should continue the discussion there.

I had the same question. Maybe this needs another offline discussion before we
act here.

I understand this concern. We could also not do this here but in issue906, at
the cost of a larger diff that would possibly be reviewed/tested only at the
next sprint (would also be fine with me).

I had a quick look. My only question is whether this change actually goes into
the direction we want. I'm not saying it doesn't, but I'm not sure if we already
have a clear plan what the eventual data organization between these classes
should be. So I want to avoid a scenario where we change things back and forth.

I'll have a look.

I opened a pull-request here:
https://bitbucket.org/SilvanS/fd-dev/pull-requests/47/issue911/diff

As another preparation for issue906, where we would like to keep PDBs and their
patterns at the same place in PatternCollectionInformation, using
PatternInformation objects, we should get rid of task proxies that are currently
stored there just for commodity. Since this induces a diff at many files, I'd
like to do this here in a separate step.