On 05/15/2018 03:42 AM, Jürgen Spitzmüller wrote: > Am Montag, den 14.05.2018, 13:51 -0400 schrieb Richard Kimberly Heck: >> So probably 2d6173d8103 was a mistake: We should just have said that >> you >> can't do that. But that was before the includeonly support, I >> believe, >> so maybe it made sense then? I'm not sure. > Note that it does not work anyway. Since we do what we can do dis- > couple such a non-output parent in the latex output routine (via > ignore_parent), the math macros are in fact _not_ inherited in > standalone situations! Since nobody complained about that during the > last years, I wonder whether this is really such a widespread usage. > > Anyway, currently our code contradicts itself: clone tries to include > non-relevant parents, and the latex routine does everything it possibly > can to revert that -- with a mixed result. > > This is a betwixt-and-between situation that does not help anybody, and > leads to bugs that are hard to identify (such as this one). > > So can we please, please, cleanly separate those buffer from the start?
I take it that your patch essentially reverts the one I mentioned above. I'm happy for it to be reverted at this point. As far as the larger issues you mentioned are concerned, let's ponder those. Riki
