> may be best solution is to alter all text fields to text type.

I have been thinking about this kind of solution, but I consider it
quite a risky activity since it may lead to a possible DoS attack
against the database server. Correct me if I'm wrong, but I see it so
that any person would be able to provide a META Description of up to
65536 bytes and such a long description would be silently added to the
database. By creating more pages in the indexed webspace, one could
fill up the disk on our server (though there's still >40GB left at
the moment ;) or at least make the search engine perform really,
really slow.

>From my experience as a web designer I can claim that well-formed HTML
document's META Description shouldn't exceed, say, 1024 bytes, thus I
consider limiting the size of META Description a good idea.

I see three solutions to the problem:

1st & IMHO better - the length of inserted strings should be adjusted
on mnogoSearch's side, eg. via the way of strncpy(src,dest,length)
before executing a query. This is a best solution because no redundant
data (which will be dropped later on) are transmitted over the
connection to the database.

2nd & also acceptable - if Postgres finds out that indexer tried to
insert, say, 1024-character string to a field, which has a limit of
512 chars, additional data are dropped and the query is successful.

3rd & less acceptable - if indexer encounters a page with too long
META Description (as I mentioned, the problem applies to other
fields in the url table as well), it executes a query like it does
now, the query fails (thus no page is being added to the database),
but indexer doesn't terminate, but - instead - keeps working.

I'd be grateful if you told me what to do to achieve any of these
effects (or to solve the problem in completely different way). It is
also  possible that my indexer's config is somehow broken, because if
it were a bug, it would affect all mnoGoSearch 3.2.3 users on
PostgreSQL, not just me :)

There was no problem like that in mnoGoSearch 3.1.19, which I used
previously, but unfortunately I decided to migrate to the newer
version ;)

Thanks for your engagement so far and looking forward for other
possible solutions,

 :. :. :: Piotr Szczepanski
 ::.::.:: www.piotr-szczepanski.org
 :' :' :: RLU#178869 | PS486-GANDI | Staszic.waw.pl - Webmaster


___________________________________________
If you want to unsubscribe send "unsubscribe general"
to [EMAIL PROTECTED]

Reply via email to