Subject: Re: [PATCHES] Infrastructure changes for recovery
From: Simon Riggs <simon@2ndQuadrant.com>
To: Andrew Dunstan <andrew@dunslane.net>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers <pgsql-hackers@postgresql.org>
In-Reply-To: <48E5544B.5050103@dunslane.net>
References: <1218109459.4549.423.camel@ebony.2ndQuadrant>
	<1220213074.4371.173.camel@ebony.2ndQuadrant>
	<1905.1220895241@sss.pgh.pa.us>
	<1221080797.3913.768.camel@ebony.2ndQuadrant>
	<15776.1221594318@sss.pgh.pa.us>
	<1221725145.3913.2315.camel@ebony.2ndQuadrant>
	<16992.1221743196@sss.pgh.pa.us>
	<1221745847.3913.2408.camel@ebony.2ndQuadrant>
	<18549.1221746983@sss.pgh.pa.us>
	<1222121161.4445.248.camel@ebony.2ndQuadrant>
	<1222184541.4445.400.camel@ebony.2ndQuadrant>
	<28973.1222381719@sss.pgh.pa.us>
	<1222985488.4445.1625.camel@ebony.2ndQuadrant>
	<48E5544B.5050103@dunslane.net>
Content-Type: text/plain
Date: Tue, 07 Oct 2008 09:17:25 +0100
Message-Id: <1223367445.4747.105.camel@ebony.2ndQuadrant>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit


On Thu, 2008-10-02 at 19:07 -0400, Andrew Dunstan wrote:

> Simon Riggs wrote:

> > Just seen this patch has been bounced into November CommitFest, even
> > though the new patch fixes all of the concerns raised.
> >
> > I'm concerned that this is going to make the final Hot Standby patch
> > fairly large, which will make it even harder to review, test and
> > generally get accepted.
> >
> > What's the best way to make this easier for you/others to review?

> The fact that it's been put on the November list doesn't mean it can't 
> be reviewed and committed before then.

But that seems unlikely to be the case.

A patch specifically marked as "required for other work" has been
delayed by more than 5 weeks on queue and nobody was ever assigned to
review it. That was exactly the problem CommitFests were supposed to
resolve and from my perspective this is a systemic failure. If I had
submitted the patch a month late it wouldn't have got reviewed any
earlier, yet many people would cry foul (why?). The current system means
I have to code up to the deadline, officially do nothing for a month,
then respond within hours to code reviews or have the patch rejected for
another month. It works great for minor patches, but its simply not
working for bigger features. It's just not possible to be fully
available to respond to reviews, yet at the same time not able to work
more than about 25% of the development calendar.

Luckily Tom reviewed it, but with no commit and no guidance on how to
proceed this still leaves me in a difficult position.

I'm forced now to leave much of this code behind, since I cannot now
complete Hot Standby at the same time as having bgwriter active during
recovery, if that code is at risk of causing the whole thing to be
rejected. Are the two together a risk? No. Is developing them together
harder? Yes. Do *I* trust my own code? Yes, but its reviewers that
count. Is it a good thing for Postgres to leave this code behind?
Probably not. Can I add it later? Maybe.

-- 
 Simon Riggs           www.2ndQuadrant.com
 PostgreSQL Training, Services and Support