Add assertions for the results of CHANGESET_REQ

[marcelklehr]

fails nearly all the time :/
The assertions are most likely not broken, so I think the code is.

[marcelklehr]

don't pull this, of course, just checkout the branch and test

[marcelklehr]

heh, the assertions are most likely broken. That stuff works completely counter-intuitive :/

[marcelklehr]

Now, it works with the current tiemslider, but fails when used with #2040...

So, CHANGESET_REQ uses different offsets for forward changesets and backwards cahngesets, this might be because there is an internal revision count that counts '\n' as rev 0 and an external rev count that uses the preset pad contents as rev 0.

I'm getting tired of trying to wrap my mind around this, right now. The assertions code probably needs to load another revision for the forward offset, god knows what else might be hidden in teh codez if that still has no effect.

[marcelklehr]

solution was to treat CHANGESET_REQUEST({start: X}) not as a revision number, but as a changeset number (including the changeset that adds the default pad text.)

Sadly rev0 (which would equal \n, for all pads) is not available, instead, rev0 (as stored in the db) is the result of applying cs0 to \n, so that's kinda inconsistent, here.

[JohnMcLear]

@marcelklehr what is the point in this pull request? :)

[marcelklehr]

well, I tried to wrap my mind around the revision count(s) while working on the timeslider++ branch. closed