in reply to Re^2: Postgresql and space characters
in thread Postgresql and space characters
Now that you've shown an example, it does seem very strange that P-SQL's "order by" behaves like that. If you are using a perl script for both storing to and fetching from P-SQL, you might try replacing spaces with a control character -- e.g. do tr/ /\x1a/ on the strings before loading the data into the table, and then tr/\x1a/ / on the strings after you pull them back out from the table.
But I don't have a P-SQL system available to test this on... maybe it would ignore control characters the same way it ignores spaces. If so, you have to figure out whether you can use something like "/" or "*" or "#" or "%" as a replacement for spaces -- whatever will "stick" during the P-SQL sorting, but is not already present in the original data.
Is there too much of this data to be held in memory at one time? If the perl script that loads the table has all the data in one array or whatever, you could create your table like this:
create table s ( a varchar(255), o int ); # insert the string and its numeric sort order as set by perl's sort select a from s order by o;
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re^4: Postgresql and space characters
by BerntB (Deacon) on Sep 18, 2006 at 09:27 UTC |