Re: Re: finding inplace alters
Thanks Andreas. I did something similar:
SELECT
s.hexdata[1,8]
INTO
v_hexdata
FROM
sysmaster:syssltdat s
WHERE
s.partnum = '0x100001' AND
s.pagenum = 1 AND
s.slotnum = 1 AND
s.slotoff = 0;
IF v_hexdata = '01001000'
THEN
LET v_endian = 'LITTLE';
ELSE
IF v_hexdata = '00100001'
THEN
LET v_endian = 'BIG';
ELSE
RAISE EXCEPTION -746, 0, 'Invalid Endianess calculation...
Check procedure code!!!';
END IF
END IF
Tested on little endian and apparently it works.
If somebody can test it on a Big Endian machine (Solaris, AIX, HP-UX
Itanium....) I'd appreciate...
Just run the query and see if it returns '00100001'
Kind regards.
P.S.: Pending IPA finder through SQL looks good (although the code is a bit
odd, it is working).
I'll post it as soon as I test it on a "real" system to check the running
time... For now, just on my VM which is just a "toy"...
On Wed, Jan 12, 2011 at 10:10 AM, Andreas Legner
<andreas.legner@de.ibm.com>wrote:
>
> Hi Fernando,
>
> you mean how to determine engine's endianess using an SQL query?
>
> There might be more elegant ways, but what about
>
> select first 1 hexdata from sysmaster:sysrawdsk;>
> This would return first 16 raw bytes from root chunk, so something starting
> on "00000000 0001" from big endian and "00000000 0100" from little endian
> servers.
> Of course this is restricted to user informix.
>
> In case this page zero uses an old page header (pre-9.40 page), you'd get
> "00000001" or "01000000" respectively. You could tell the two cases by
> checking if first 8 nibbles all are zeroes.
>
> More easily, you could:
>
> select first 1 hexdata[1,4] from sysshmem;>
> This should return first shmem segments magic number 0xe0b7, as either
> "e0b7" or "b7e0".
>
> Cheers,
> Andreas
>
>
>
>
> From:
> Fernando Nunes <domusonline@gmail.com>
> To:
> Andreas Legner/Germany/IBM@IBMDE
> Cc: informix-list@iiug.org Date: 12.01.2011 03:06 Subject: Re: Re: finding
> inplace alters
> ------------------------------
>
>
>
> Andreas,
>
> Thanks for the reply, and don't bother... You were right... A checkpoint is
> needed.
> So, at this moment I have something that apparently works... But needs some
> refinement and further testing...
> Also... I need a way to find out if the platform is little endian or big
> endian.... any ideas?
>
> Regards!
>
--
Fernando Nunes
Portugal
http://informix-technology.blogspot.com
My email works... but I don't check it frequently...