Re: Re: Re: Heap sorting in perl

[pg]

If those are rows in a database, I agree with dws, just let database do it.

If that particular column is not indexed, I suggest index it, as you need it.

"order by" is usually not too expansive, especially when it only involves indexed range scan.

Depends on your database, some database directly support syntax that allows to retrieve first N rows.

[dws]

If that particular column is not indexed, I suggest index it, as you need it.

If this is a write-mostly, query-infrequently table, indexing could significantly degrade performance. For log/audit files that record large numbers of miscellaneous value, indexing the values is rarely an effective strategy.

[pg]

When this is generally correct, I don't really see a reason why a person want to "sort by" some column containing some miscellaneous values. If unfortunately that's the case, then should look into the data, and have some sort of key information extracted.

Any way, I believe blakem has his stuff well organized, and what he is doing is reasonable.