On Fri, Jun 6, 2008 at 3:46 PM, Ethan Collins
<collins.ethans+psql@gmail.com> wrote:
> I have got a table that holds 2 columns: date_id and info. date_id is
> bigint, created as, for eg. 17 May 2008, 19:05 hrs => 200805171905. This
> table is populated with {date, info} pair that I receive frequently. AND
> there can be mltiple such entries with a single date_id but varying info,
> for eg. 20 entries with same date_id but different info. I haven't created
> any primary key or index.
> I use query that looks like this:
> SELECT date_id, info FROM netproto_entries ORDER BY date_id DESC
> LIMIT <limit_val>;
> Questions:
> - For higher values of limit_val (say 70/75/..), when I change it, the order
> in which I get the entries changes. Eg. it shows me one info at the top at
> one time, and another on the next time with a different value of
> limit_val.
> - My intent is to fetch the entries from the database in LIFO order, most
> recently entered entry at the top. For the elements with the date_id being
> the same, they are not in LIFO order. I understand that I can add another
> incrementing index column and sort the table based on that. But, without
> adding this,
> does postgresql have any method/configuration to get me output in LIFO
> order ?
How is it to tell which record is which?
> And are there any better methods available compared to adding another
> index
> column (as mentioned above) ?
No need for another index. Just create a serial column and use it to
break the ties.