I want some experience with Sphinx so that can verify the premise that
Sphinx is better than Lore. So, the experience isn't necessary for the
conversion, it's necessary for me to be comfortable that the conversion
is a good idea.
Oh. I am definitely taking that as a given. Or rather: having frequently interacted with the output of both sphinx and lore, I was already confident that it was better than Lore in most ways before we began this conversation. I had a few specific misgivings, which Kevin has since addressed.
Sphinx clearly has the *features* that we need; Kevin's analysis looks pretty thorough, and even the current error-laden converted documentation serves as a reasonable proof of concept. Plus, the plan specifically includes resolving the remaining issues with that documentation as a prerequisite to getting it merged.
So while I could believe that there will be bad things about Sphinx which may surprise me, I find it hard to believe that I could be surprised by any bug in Sphinx so awful that it would be *enough worse* than Lore to make continuing to maintain Lore a preferable option, given the presence of a documentation maintainer willing to do the drudgework of conversion.
That said, it might be a good idea to do the Divmod project conversions first for other reasons. They have less documentation, so there's less to review, and the process could be completed more quickly, letting us get some benefit out of the process earlier on. Since these projects are the only known remaining users of lore, we could get rid of lore immediately when twisted itself is converted, rather than letting it linger until its users have converted.