looking for references headers for the ^ command

looking for references headers for the ^ command

[Greg Stark]

If the NOV informatio contains a referencesentry but the actual article is
missing it Gnus seems to do somewhat unpredictable things, sometimes ^ works
sometimes it doesn't. This situation arises because my backend interfaces to a
system that implements threading but not via references headers. So i generate
synthetic message-id and references headers for the nov data but i am loath
the actually fiddle with the contents of the articles, especially since some
of them contain real message-id's and references headers and i wouldn't want
my synthetic headers to ever leak out.

I think the right thing is for gnus to try any references headers provided by
either source, the NOV data and the actual headers. I'm not sure which it
should prefer if both return message-id's for available articles, perhaps it
should be a variable the backend could tweak. 

I'll look into doing this myself since it's likely it only affects me :)
but i just though i would mention it in case someone else is concerned.

greg

Re: looking for references headers for the ^ command

[Justin Sheehy]

Greg Stark <gsstark@MIT.EDU> writes:

> I think the right thing is for gnus to try any references headers provided by
> either source, the NOV data and the actual headers.

This would defeat the entire purpose of NOV, which is to make it
unnecessary to read the entire articles.

If the NOV data is not going to be sufficient, then don't bother
providing it at all.

-Justin

 

Re: looking for references headers for the ^ command

[Greg Stark]

Justin Sheehy <dworkin@ccs.neu.edu> writes:
> This would defeat the entire purpose of NOV, which is to make it
> unnecessary to read the entire articles.
> 
> If the NOV data is not going to be sufficient, then don't bother
> providing it at all.

You misunderstand, i've looked at it now, the current logic is to use the NOV
data except when it's the current article in which case Gnus looks for the
actual fields. Going back to the original source makes sense for most
situations in which the NOV data is actually generated based on real
references headers.

What i'm asking for is for Gnus to fall back on the NOV data if it exists and
the real headers don't work. i've looked at the source and it should be small
change and it won't affect normal cases where the actual headers are valid.

greg

Re: looking for references headers for the ^ command

[Lars Magne Ingebrigtsen]

Greg Stark <gsstark@mit.edu> writes:

> What i'm asking for is for Gnus to fall back on the NOV data if it exists and
> the real headers don't work. i've looked at the source and it should be small
> change and it won't affect normal cases where the actual headers are valid.

Ok; mail me patches.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen