Yup, works for us. @Arlolra we just need to verify that we don't have any baked-in assumptions about this. I don't think we have and the parsing works fine when these are absent, but, doesn't hurt to double-check.

Considering T98995, it's probably fine. But if we decide to do that, we should also remove the functionality from Parsoid altogether.

In T170997#3450399, @Arlolra wrote:

Considering T98995, it's probably fine. But if we decide to do that, we should also remove the functionality from Parsoid altogether.

By 'to do that', I presume you mean this ticket. But, the keyword in the ticket is 'temporarily'. We have a bunch of performance work ahead of us, and this is a useful feature. Plus, incremental parsing (the holy grail) does require access to previous revision HTML.

In T170997#3450399, @Arlolra wrote:

Considering T98995, it's probably fine. But if we decide to do that, we should also remove the functionality from Parsoid altogether.

@Arlolra we can resurrect it in RESTBase at some point, just during the transition period it will greatly simplify the reasoning about what's going on. The previous revision/render should be in storage at some point, just I'd better switch this off while we're transitioning to the new storage model, because in the beginning the previous versions will not be in storage, so it will either double parsoid load (if we generate previous demand) or will just silently disappear - which I was worried could break something in parsoid.

So I guess the consensus is to remove it right now for the transition period and resurrect later after we've tested the basic functionality and populated the storage.

Sounds like I didn't read the last paragraph :/

In T170997#3450479, @Pchelolo wrote:

So I guess the consensus is to remove it right now for the transition period and resurrect later after we've tested the basic functionality and populated the storage.

Yes.

PR: https://github.com/wikimedia/restbase/pull/837