John Machin wrote:
>
> On 8/03/2009 9:02 AM, baxy77bax wrote:
>> hi,
>>
>> i need help with attaching databases. in my last post i recived some
>> valuable info on limitation of attached databases (Thanx !!). since i'm
>> using perl (its DBI) all modifications like cache_size, max page number ,
>> journal size ... i've introduced through PRAGMA but for limitations on
>> number of attached databases, there is no pragma statement :confused:
>> (or
>> al least i can't see one :-/)and i would need to some how change the
>> default
>> value from 10 to 30 databases.
>>
>> can anybody help me with that ?
>
> It is not a "default value" that can be changed at run time by using a
> pragma. You would need to get the source, change the limit, and compile
> your own version of SQLite.
>
> Could you find a use for the DETACH statement?
>
> What real-world scenario is driving your technical requirement to have
> 31 databases active simultaneously?
>
> Where in the real world does the magic number 30 come from?
>
> Will you ever need more than 30? Have you noticed that (on a 32 bit
> machine) you cannot increase the compile-time limit to more than 30?
>
> Do you actually have (or plan to have) an SQL query that actually
> mentions each of the 31 databases individually? If no such query, why do
> you assert that you need 30 databases attached simultaneously? If there
> is such a query, is it 30 SELECTs (one per attached database) joined by
> UNION?
>
> Could you achieve the real-world objective by processing the 30 attached
> databases one at a time?
>
> HTH,
>
> John
>
>
> _______________________________________________
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>
>
the reason why i need to attach 30 db 's at once is because i 've changed
the blast program(alignment program - big outputs ) to format and save the
results in the lite db. since blast could be run on a grid engine it then
creates more than 10 sqlite databases. those results then i have to merge
into one db for further processing(processing is not created for to be done
on more than 4 cores, not relevant). so some usual serial(parallel) job is
done on 30 cpu's(that is the maximum number ob cpu's that can be given to a
job (human limited) on the server i use). of course one day maybe i'll use
more than 30 cores but for now 30 is the magic number
detach statement is not the option since the search (processing) has to be
done simultaneously once the job is done. i could dynamically search the
db's pericyclically but that would demand extra disc space and would slow my
process dramatically (that is the reason why i'm pushing for 30 jobs to be
done at once)
so the only way out i see is the attaching all 30 db's at once :)
suggestions are more than welcome on how to solve this problem more
elegantly !
thnx
--
View this message in context:
http://www.nabble.com/change-the-limitation-of-attached-databases-through-perl-tp22392320p22396377.html
Sent from the SQLite mailing list archive at Nabble.com.
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
