Re: Clarification reqeusted for "select * from a huge table"

Gokulakannan Somasundaram wrote:
>>>     I also noticed that it doesn't crash with psql, but it takes a
>>> long time to show the first set of records. It takes a long time, even
>>> to quit after i pressed 'q'.
>>>    With oracle SQLPlus, it is quite instantaneous.
>> Again, you're measuring different things. What is the time to the *last*
>> row?
> 
> I made this point, because people usually fire select * from table
> query in the psql prompt to get a feel of the table.

Well, I don't. If nothing else, it's difficult to fit a whole row across 
the screen with many tables.
> Ofcourse they can
> fire select * from table limit 10; But i just feel its more key
> strokes and it would be better, if it is interactive. I accept that it
> is a debatable point and people can prefer otherwise to have more key
> strokes.

If you want few keystrokes to see what columns a table has, how about
\d table

> Imagine, you need a large batch operation. In oracle we can fire the
> SQL and we can be sure that the client won't crash, but with postgres
> we have a region of uncertainity.

Well, if your client doesn't know if it can handle 1 million rows, maybe 
it shouldn't ask for them?
> There are some JDBC hints like
> setFetchSize(), which actually affects Oracle's behaviour. But it
> doesn't seem to do anything with postgres. But JDBC has declared these
> commands as hint commands and has provided a warning to users, about
> the fact that it may get ignored

I see you've checked and seen it does work. I'm not a java man, but I do 
know there are ways to control this. Perhaps try the -jdbc mailing list.

In any case, I think we've established it's nothing for the hackers list.

--   Richard Huxton  Archonet Ltd