[ Sorry for not getting in on the fun earlier, but I'm not subscribed to
pgsql-ports, so I had no idea this discussion was going on over there. ]
Bruce wrote yesterday:
> My recommendation(hold on to your seats) is to take the current cvs
> tree, patch it with Uncle George's patches and any others needed, and
> release a 6.5.2 release that addresses alpha. We can back-patch 6.5.2,
> but there is really no reason to do that. There is really nothing
> 'special' in the current tree. In fact, the most risky of them are the
> alpha ones,
I have to disagree. I've already committed a ton of parser and optimizer
changes, which are far too unproven to call 6.5.2. If there is a 6.5.2
it *must* be as small as possible a diff from 6.5.1, not something based
on the current tree.
I have not seen the proposed Alpha patches, but I gather they include
diffs to get rid of the "fmgr calls not passing char and short
parameters properly" problem, which we know also exists on PPC and
perhaps some other architectures. That strikes me as being a fairly
major problem that can't really be fixed in 6.5.* --- the diffs would
certainly be extensive, and would they have gotten enough testing to
risk being put into a patch release? I certainly would not trust
them without testing on multiple architectures...
(BTW, you may recall that I have a proposal on the table to fix the
parameter problem via a wholesale revision of the fmgr interface.
If we go that route, uglifying the code by changing chars and shorts
to Datum will be work that'll have to be undone later.)
regards, tom lane