Re: nnvirtual groups and article warping

[Andrew Cohen <acohen@ust.hk>]

>>>>> "ESF" == Eric S Fraga <e.fraga@ucl.ac.uk> writes:

    ESF> On Sunday,  6 Mar 2022 at 21:04, Andrew Cohen wrote:
    >> That is a bug (at least for nnselect). I use this constantly in
    >> nnselect and it works fine. We should try to track this down; if
    >> I had to guess it has something to do with the search engine you
    >> are using---I can't recall but think it is notmuch?

    ESF> Sorry, maybe I confused things: A W works just fine with
    ESF> notmuch searches and nnselect.  I was commenting on the need
    ESF> for A W for nnvirtual groups to allow me to see a whole tread
    ESF> using A T (which doesn't work in an nnvirtual group); the
    ESF> question I was answering was what the use case for A W would be
    ESF> if marks etc. were all propagated back to the actual groups.

Got it, thanks. So the use is to get the functionality of A T when using
nnvirtual, but not anything else. Thats good, since I think I have
solved the remaining issues with emulating nnvirtual with nnselect (in
which case A T works correctly). In the end its all pretty trivial
except for managing the combined group information (nnvirtual doesn't
store any of it, while nnselect does---the information is pretty large
because of the interleaving of groups and would cause the newsrc size to
balloon if we didn't do something; this is the only change needed to
nnselect to make it work).  My push rights are in limbo over copyright
confusion, but once that is settled I'll push it and get some
testing. It should work as an exact drop-in replacement, but we will see
if anything is missing. (And I still encourage the alternate paradigm I
previously pushed for, which I think would work much better for you :))

Best,
Andy
-- 
Andrew Cohen