Re: relcache refcount - Mailing list pgsql-hackers

From Zeugswetter Andreas SB SD
Subject Re: relcache refcount
Date
Msg-id 46C15C39FEB2C44BA555E356FBCD6FA40184D0D4@m0114.s-mxs.net
Whole thread Raw
In response to relcache refcount  (Alvaro Herrera <alvherre@dcc.uchile.cl>)
Responses Re: relcache refcount  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
> BTW, what are your plans for state saving/reversion for the lock manager
> and buffer manager?  The lock state, in particular, makes these other
> problems look trivial by comparison.

Why can't we keep all locks until main tx end ? Locks are not self conflicting
are they ? So the only reason to free them would be to improve concurrency,
and imho we don't need that. I guess I am just not seeing this correctly.
(I am assuming that a deadlock will still break the whole tx)

Andreas


pgsql-hackers by date:

Previous
From: Gaetano Mendola
Date:
Subject: Re: pg_begintypend
Next
From: pgsql@mohawksoft.com
Date:
Subject: Re: [pgsql-hackers-win32] Sync vs. fsync during