Re: Merge compact/non compact commits, make aborts dynamically sized

On 2015-02-27 16:26:08 +0900, Michael Paquier wrote:
> On Wed, Feb 25, 2015 at 8:10 PM, Andres Freund <andres(at)2ndquadrant(dot)com> wrote:
> > On 2015-02-24 20:51:42 +0200, Heikki Linnakangas wrote:
> >> On 02/20/2015 05:21 PM, Andres Freund wrote:
> >> >There's one bit that I'm not so sure about though: To avoid duplication
> >> >I've added Parse(Commit/Abort)Record(), but unfortunately that has to be
> >> >available both in front and backend code - so it's currently living in
> >> >xactdesc.c. I think we can live with that, but it's certainly not
> >> >pretty.
> >>
> >> Yeah, that's ugly. Why does frontend code need that? The old format
> >> isn't exactly trivial for frontend code to decode either.
> >
> > pg_xlogdump outputs subxacts and such; I don't forsee other
> > usages. Sure, we could copy the code around, but I think that's worse
> > than having it in xactdesc.c. Needs a comment explaining why it's there
> > if I haven't added one already.
>
> FWIW, I think they would live better in frontend code for client applications.

What do you mean with that? You mean you'd rather see a copy in
pg_xlogdump somewhere? How would you trigger that being used instead of
the normal description routine?

> +/* free opcode 0x70 */
> +
> +#define XLOG_XACT_OPMASK 0x70
> There is a contradiction here, 0x70 is not free.

The above is a mask, not an opcode - so I don't think it's a
contraction. I'll expand on the commentary on the next version.

Thanks for having a look!

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services