bug reporting: reproducible test case
you either have a reproducible test case, or you don't. if the issue cannot be reproduced, it does not exist.
next.
-bdbafh
I think that the parent hit on something here.
Lots of issues are "Fixed in 10.2" and may be waiting out there in the "10.1.0.5" patchset that is due to ship in Jan 2006.
Its the backporting the one-off patches (not patchsets) to a current release that cause fear of ending up in Opatch hell having to file yet another iTAR with Oracle Support.
-yet another patching fool
23 years ago, I used to do the same thing.
Solarization.
Posterization.
take a solarized high contrast film "print" and turn it into a line drawing.
then again, that was in my parent's basement.
http://www.eff.org/IP/DMCA/US_v_Elcomsoft/us_v_skl yarov_faq.html
this is the same Adobe that went after Dmitri Sklyarov.
what has changed in their org, board of directors, corporate mindset that should convince me to ever forgive them?
Yep - I didn't think so.
Lets have a little contest as to how large of objects we can shove up the collective rectum of Adobe.
-me
and turn them on as the authorities see fit... no court order required.
time to go look at EFF and EPIC sites again.
maybe this time I'll actually donate like I've meant to before...
your post was worth reading until...
"2 - no commercial software is written for it."
"not much" - agreed.
"much less" - agreed.
"No" ?? - you lost me there.
Whomever modded your post up with that piece of shit comment in it is an udder fucking moron.
and literate windows users that have set IE to run ActiveX controls don't get the chance to decide.
fuckwad.
The windows servers don't have sshd running?
Well then get it installed, dammit!!!
we've got cygwin on our win32 servers and an sshd running, but I'll admit that we're only authenticating against local accounts, not domain accounts (kinda stealthy).
-bdbafh
yeah, I've got one of those boxes also.
still running 9.2.0.3 on a RH8 install.
only ports open are ssh and ssl, tns listener running on loopback only. gotta hit apache on it to get in, audited out the wazoo.
can't wait to deprecate that box for Apache 2 / 10g / RHEL 3.0 ES update 4 / (something faster than a P II 350)
so pickup a cygwin CD and run the xserver on your win32 laptop.
I've installed 9.2 and 10.1 on RHEL 3.0 ES via the Cygwin Xserver on the WinXP boot on me ole laptop.
Its covered at sites like:
http://www.puschitz.com/
so create the views FORCE and recompile afterwards.
yes, 8.1.6 lacked a good dependency-driven utility such as utlrp, so you had to recompile things by hand at times.
you upgraded to 9.2 and got over it.
10g Release 2 is soon upon us.
in Oracle, one would use the RETURNING clause to get the sequence value used to populate the column.
Or did you want to encur more SQL*Net round trips just to get back.currval?
of course you didn't.
do you have any idea as far as how many security vulnerabilities exist for oracle 8.1.7.2?
Dude, I hope that the only ip address that the listener is configured for is 127.0.0.1.
loadphp can crash your deal, and allow for takeover of the system.
several iterations of remote vulnerabilities in the listener.
several security alerts on the oracle executable.
http://www.oracle.com/technology/deploy/security/a lerts.htm
upgrade or die.
8i R3 is now desupported.
logon to metalink.
click on support/availability.
enter your platform/version info.
eureka! 8.1.7 is obsolete.
pick one:
9.2.0.x
10.1.0.x
wait for 10.2
upgrade or die.
SQL> set autotrace on explain statistics timing on
or
SQL> exec dbms_support.start_trace(binds=>false, waits=>true);
and examine the trace file, hit it with tkprof.
your (wrong) opinion will change.
methinks that your statistics are stale, or non-existent.
all of the oracle docs are up on otn.oracle.com.
read them.
where 1=1
chew on that a bit.
if one is going to *maintain* an aggregate, that would imply locking the base table in order to get a consistent read of that entity. oh, you're going to maintain the SCN of when the aggregate was calculated and the difference from that? as of what point?
talk about a race condition.
as others have posted, the optimizer should use an index, PK first, that has not nulls in it to determine the row_count for the table, not something that is maintaining a row_count in a multiuser system.
Oracle Support will bail your arse out.
Provided that your support contract is paid up, and sometimes, even if it is not - you will be able to get your data back out even of a corrupted database.
I have only seen 2 corrupted databases, due to massive hardware failures (degraded over time).
six grand USD for Oracle Standard Edition One, about a grand a year for maintenance (upgrades) and support - and this big, bad, expensive RDBMS seems quite inexpensive.
A properly configured database can run unattended for 3 months at a time, provided that it has been setup correctly, and someone is grepping the logs.
(quarterly security patchsets are now the norm).
At $15K USD per CPU, I'm interested in Postgres.
At $5K USD per CPU, applications run on Oracle.
bdbafh
Slashdot Mirror
bug reporting: reproducible test case you either have a reproducible test case, or you don't. if the issue cannot be reproduced, it does not exist. next. -bdbafh
I think that the parent hit on something here. Lots of issues are "Fixed in 10.2" and may be waiting out there in the "10.1.0.5" patchset that is due to ship in Jan 2006. Its the backporting the one-off patches (not patchsets) to a current release that cause fear of ending up in Opatch hell having to file yet another iTAR with Oracle Support. -yet another patching fool
23 years ago, I used to do the same thing. Solarization. Posterization. take a solarized high contrast film "print" and turn it into a line drawing. then again, that was in my parent's basement.
Horse. Barn door. Genie. Bottle. Pandora. Box. Would even a fine of $100 per account make a dent in their bottom line?
http://www.eff.org/IP/DMCA/US_v_Elcomsoft/us_v_skl yarov_faq.html
this is the same Adobe that went after Dmitri Sklyarov.
what has changed in their org, board of directors, corporate mindset that should convince me to ever forgive them?
Yep - I didn't think so.
Lets have a little contest as to how large of objects we can shove up the collective rectum of Adobe.
-me
meant to post anonymously. damn you, autologin.
aliens stole my buick mr. T ate my balls
and turn them on as the authorities see fit ... no court order required.
time to go look at EFF and EPIC sites again.
maybe this time I'll actually donate like I've meant to before ...
your post was worth reading until ...
"2 - no commercial software is written for it."
"not much" - agreed.
"much less" - agreed.
"No" ?? - you lost me there.
Whomever modded your post up with that piece of shit comment in it is an udder fucking moron.
and literate windows users that have set IE to run ActiveX controls don't get the chance to decide.
fuckwad.
The windows servers don't have sshd running? Well then get it installed, dammit!!! we've got cygwin on our win32 servers and an sshd running, but I'll admit that we're only authenticating against local accounts, not domain accounts (kinda stealthy). -bdbafh
yeah, I've got one of those boxes also. still running 9.2.0.3 on a RH8 install. only ports open are ssh and ssl, tns listener running on loopback only. gotta hit apache on it to get in, audited out the wazoo. can't wait to deprecate that box for Apache 2 / 10g / RHEL 3.0 ES update 4 / (something faster than a P II 350)
so pickup a cygwin CD and run the xserver on your win32 laptop. I've installed 9.2 and 10.1 on RHEL 3.0 ES via the Cygwin Xserver on the WinXP boot on me ole laptop. Its covered at sites like: http://www.puschitz.com/
so create the views FORCE and recompile afterwards. yes, 8.1.6 lacked a good dependency-driven utility such as utlrp, so you had to recompile things by hand at times. you upgraded to 9.2 and got over it. 10g Release 2 is soon upon us.
in Oracle, one would use the RETURNING clause to get the sequence value used to populate the column. Or did you want to encur more SQL*Net round trips just to get back .currval?
of course you didn't.
Best post on this entire thread. Give that poster a copy of any of Jonathan Lewis' books.
do you have any idea as far as how many security vulnerabilities exist for oracle 8.1.7.2? Dude, I hope that the only ip address that the listener is configured for is 127.0.0.1. loadphp can crash your deal, and allow for takeover of the system. several iterations of remote vulnerabilities in the listener. several security alerts on the oracle executable. http://www.oracle.com/technology/deploy/security/a lerts.htm
upgrade or die. 8i R3 is now desupported. logon to metalink. click on support/availability. enter your platform/version info. eureka! 8.1.7 is obsolete. pick one: 9.2.0.x 10.1.0.x wait for 10.2 upgrade or die.
SQL> set autotrace on explain statistics timing on or SQL> exec dbms_support.start_trace(binds=>false, waits=>true); and examine the trace file, hit it with tkprof. your (wrong) opinion will change. methinks that your statistics are stale, or non-existent. all of the oracle docs are up on otn.oracle.com. read them.
where 1=1 chew on that a bit. if one is going to *maintain* an aggregate, that would imply locking the base table in order to get a consistent read of that entity. oh, you're going to maintain the SCN of when the aggregate was calculated and the difference from that? as of what point? talk about a race condition. as others have posted, the optimizer should use an index, PK first, that has not nulls in it to determine the row_count for the table, not something that is maintaining a row_count in a multiuser system.
use bind variables, you incompetent clod!
Oracle Support will bail your arse out. Provided that your support contract is paid up, and sometimes, even if it is not - you will be able to get your data back out even of a corrupted database. I have only seen 2 corrupted databases, due to massive hardware failures (degraded over time). six grand USD for Oracle Standard Edition One, about a grand a year for maintenance (upgrades) and support - and this big, bad, expensive RDBMS seems quite inexpensive. A properly configured database can run unattended for 3 months at a time, provided that it has been setup correctly, and someone is grepping the logs. (quarterly security patchsets are now the norm). At $15K USD per CPU, I'm interested in Postgres. At $5K USD per CPU, applications run on Oracle. bdbafh
Oracle. read the fine readme. download patch uncompress shutdown apply patchset rebuild dictionary $sqlplus /nolog
SQL> spool catpatch.lst
SQL> connect / as sysdba
SQL> shutdown immediate
SQL> startup migrate
SQL> @?/rdbms/admin/catpatch.sql;
SQL> shutdown immediate
SQL> startup restrict
SQL> @?/rdbms/admin/utlrp;
SQL> alter system disable restricted session;
SQL> spool off
SQL> quit
done.
profit.