Re: trees, trees, trees
From: geos <geos_at_nowhere.invalid>
Date: Mon, 25 Jun 2012 23:25:18 +0200
Message-ID: <jsal1q$2ae$1_at_news.task.gda.pl>
ddf wrote:
> On Jun 24, 5:46 pm, geos <g..._at_nowhere.invalid> wrote:
>
> Can you, and have you tried since posting this originally?
>
> David Fitzjarrell
Date: Mon, 25 Jun 2012 23:25:18 +0200
Message-ID: <jsal1q$2ae$1_at_news.task.gda.pl>
ddf wrote:
> On Jun 24, 5:46 pm, geos <g..._at_nowhere.invalid> wrote:
>> so, can you solve the problem with pure SQL? >> >> thank you, >> geos
>
> Can you, and have you tried since posting this originally?
>
> David Fitzjarrell
yes, I have the solution already. is is quite simple. it works with client data but I'm not 100% sure it's doesn't have any weak points though. I am still testing it. I will post the code after I clean it up along with some request for suggestions about improving its performance (though I'm not convinced if this makes any sense here).
for short: I used sys_connect_by_path on pre-prepared set of original data, regexp_replace to remove duplicate nodes (it improves performance), and xmlcast for splitting the paths (instead of level/dual trick, it took too long).
can you solve this problem with pure sql? do you know some equivalent non-hierarchical approach that works?
thank you,
geos
Received on Mon Jun 25 2012 - 16:25:18 CDT