Re: Slow count(*) again... - Mailing list pgsql-performance

On 2010-10-14 21:56, Robert Haas wrote:
> On Thu, Oct 14, 2010 at 12:22 AM, mark<dvlhntr@gmail.com>  wrote:
>
>> Could this be an interesting test use of https://www.fossexperts.com/ ?
>>
>> 'Community' driven proposal - multiple people / orgs agree to pay various
>> portions? Maybe with multiple funders a reasonable target fund amount could
>> be reached.
>>
>> Just throwing around ideas here.
>>
> This is a bit off-topic, but as of now, they're only accepting
> proposals for projects to be performed by CommandPrompt itself.  So
> that doesn't help me much (note the sig).
>
> But in theory it's a good idea.  Of course, when and if they open it
> up, then what?  If more than one developer or company is interested in
> a project, who determines who gets to do the work and get paid for it?
>   If that determination is made by CommandPrompt itself, or if it's
> just a free-for-all to see who can get their name on the patch that
> ends up being committed, it's going to be hard to get other
> people/companies to take it very seriously.
>
Couldnt you open up a dialog about it?
> Another problem is that even when they do open it up, they apparently
> intend to charge 7.5 - 15% of the contract value as a finder's fee.
> That's a lot of money.  For a $100 project it's totally reasonable,
> but for a $10,000 project it's far more expensive than the value of
> the service they're providing can justify.  (Let's not even talk about
> a $100,000 project.)
>

Hi Robert.

I can definately see your arguments, but you failed to describe
a "better" way?

Many of us rely heavily on PostgreSQL and would
like to get "this feature", but sponsoring it all alone does not seem
like a viable option (just a guess), taken into consideration we dont
even have an estimate about how big it is, but I saw the estimate of
15K USD of the "ALTER column position" description.. and the
visibillity map is most likely in the "same ballpark" (from my
perspective).

So in order to get something like a visibillity map (insert your
favorite big
feature here), you have the option:

* Sponsor it all by yourself. (where its most likely going to be too big,
    or if it is the center of your applictions, then you definitely turn
to a
    RDBMS that has supported it for longer times, if you can).
* Wait for someone else to sponsor it all by them selves. (that happens
   occationally, but for particular features is it hard to see when and
what,
   and the actual sponsor would still have the dilemma in the first point).
* Hack it yourselves (many of us dont have time neither skills to do it, and
   my employer actually wants me to focus on the stuff that brings most
direct
   value for my time, which is a category hacking PG does not fall into
when the
   business is about something totally else).
* A kind of microsponsoring like above?
* Your proposal in here?

To me.. the 4'th bullet point looks like the most viable so far..

To be honest, if it is EDB, Redpill, Command Prompt, 2nd Quadrant or
whoever end up doing the job is, seen from this perspective not
important, just it ends in the hands of someone "capable" of doing
it. ... allthougth Heikki has done some work on this task allready.

Preferrably I would like to get it coordinated by the PG project itself.
But
I can see that it is really hard to do that kind of stuff. And you would
still
face the challenge about who should end up doing the thing.

Jesper .. dropped Joshua Drake on CC, he might have given all of this some
seconds of thought allready.

--
Jesper