is there a way to convert a collection to a row?
i have a 'table of record' type
and i have to convert it to rows like those provided on select statements
why is that this channel here at all?
oracle is a fu*king sh*t

Snagnever: what exactly is the problem..?

well, the problem NOW is not exactly with oracle
but wit its PHP api
i have to use a type defined inside a collection

okay...well, don't throw a fit over Oracle, then :P

but the api simply says it doesnt exist

You got some code we can look at?

yes
let me paste on phpfi
thanks for helping me [=

Snagnever: np..I might afk on and off b/c I'm at FOSSCon, currently

ganderson, http://pastebin.org/344167
so, i can find nothing about that on google
even oracle's site ignores that
http://www.oracle.com/technology/pub/articles/oracle_php_cookbook/index.html looking here

Snagnever: what kind of error does that line produce?

arning: oci_new_collection() [function.oci-new-collection]: OCI-22303: type ""."Classificacao.InfoGrupo" not found in C:\www\LabBD2010\index.php on line 35

hmm

so
what i think i could to
is transform the collection
in a set of rows
like in a common select

Snagnever: from what I'm finding...it sounds like it's failing to get this collection from the DB, right?

yes!

http://forums.oracle.com/forums/thread.jspa?threadID=365039 <-- there's mention here (and the php API) says you can specify the schema

let me see

so maybe try specifying the schema in all capital letters
e.g. $valores = oci_new_collection($conn, 'Classificacao.InfoGrupo', 'SCHEMA')

wait
what would be my schema?
as labbd is my user
"labbd" ?

yea
put it in all caps, though --> 'LABBD'

let me see
ganderson, same error
but

hmm

if i define a type in the schema
CREATE OR REPLACE TYPE TestType is table of varchar(100)
it works

ah...that may work

but not that inside de collection

honestly, I haven't created a type in oracle before, so I'm not completely sure about that

so, it seems it's an api limitation

it may be

but i have to finish
my college work
today :(

aww :/

and i'm not in mood to move to java

have you checked some forums?

i'm not confortable with swing api
yes!
there is only one question like mine

did you post your question? Someone else may be able to help you off of there

one answere
i've tried here
only you came to help me ;x

well, if you sit tight, someone else may be able to help

ganderson,
but

you could also check if there's another channel for php programming that might have something

would you know a way
to convert the collection to rows?
it would be a solution

hmm

Fix.Convert( Collection );

what do you mean by 'rows'?
an array?

no, like the default result
in a SELECT query

hmm...I'm not sure I'm completely understanding your question, then

SELECT * FROM (ACollection) like that

okay
and you're just trying to display that resultset via PHP?

YES
yes*
sorry for the caps

well why not just execute a statement with that query in it and run through the result set, instead of using the collection api

yes
i know

okay

but the work says explictly
to use collections

oh

it's a fcking college work \=

hmm...well I'm not completely up to speed on working with the collections stuff with PHP and oracle. Unfortunately I gotta get going too, so good luck and if you find the solution, don't forget to post it on that forum you found!

ook dude
thanks for trying to help man!
:D
see you!

hiyas. i just ran a little test (mostly using the system tables) against my local XE. i wind up with ~2-3 queries per second. could that be an auto-commit problem, or is there something oyu have to tune before working with XE?

question about fractured block upon server crash
say you have a 32k blocksize, and 8k OS page size, with 512byte disk IO size
now say you have a table with rows that are 2k in size
if you insert say 15 rows into the block, then delete 13 of those

i have a question:

you are forgetting alot of other things kpanic99

why define types and exceptions while apis cannot use them?

i'm aware of redo replay

yeah and not just that
but it seems like you are trying to do homework or something

but if an insertion MOVEs an old row closer to the head, there is no redo change vector

because theory isnt real life

the question:
the undo entry that gets saved prior to flushing block io...
is it the entire block, or just the before image of the affected ROW:?

many many more

why wouldn't you set your db blocks to 8K as well? that way every oracle i/o request the OS only needs to retrieve one block

a database wouldnt be very useful if it worked with just 1 block at a time

there are many reasons to reduce the number of rows in a block
RAC is one reason
fewer rows means less contention, less block shipping

god i wish i only worked in theory
life would be so easy

fractured blocks are very rare kpanic99
are you actually having this problem

i can give you a real life repro rjarett if it will get you to be quiet

lol
dude stfu
ask your teacher

ok

This isnt real life

i'm a 18 year veteran of clustering, came here to ask the 'experts' about oracle

(Action) cuts self I AM ALWAYS DISAPPOINTING RJARETT

not you zasz

yesssss

anybody here that can find their anus with both hands?
(Action) that was directed squarely at rjarett for the innocent onlookers

at least mine doesn't have a di*k in it like yours
o

either you guys are willing to carry on a technical conversation, or you're idiots and i should just move on
looking for someone with a clue, rather than someone who's sole purpose is to flood the channel with insults

eh. ex oracle employee who gets pissed at theoretical "homework" style questions which arent even valid to how a system works

"how a system works"

ask your undo question again

i have worked many fractured block analysis escalations for clients
does the undo store the entire block, or just the before image of the change vector?

blocks i believe
ill double check

i always thought it was the whole block, but after doing some digging i'm doubting my prior understanding
grazie zasz

partial except the first time it is changed while in backup mode.

the concern here is if a row gets moved up closer to the header as a result of shrinking of a row above it, due to an insertion -- the row could wind up lost with larger blocksizes
talking about normal runtime, not online backup mode (which stores the writes in redo instead of flushing to datafiles)
i've worked my share of "unexplained" torn/fractured blocks, and this little revelation might help explain things

RJarett: wouldn't before images of the entire block be written to UNDO?
since UNDO tablespaces are composed of undo segments (which are in turn composed of extents and blocks)

asking this another way, might you have some tidbit for analyzing undo datablocks themselves

kpanic have you thought of opening an SR with oracle

od

yeah, but lately those don't get very far
hit/miss if the data that comes back is accurate
another angle to consider here, any way to dump the current undo tablespace structures for analysis similar to mining redo log change vectors
thinking of writing a testcase to heavily exercise this scenario, crash the box, and then go shopping for a fractured block with dbv, once identified then look for affected rowid's in redo dump
but to be complete, would need a method to dump the undo as well which I've never done

such is beyond me but let me know what you find :)

fair enough, i'll ping the results here

hi! has anyone worked with PHP + Oracle with OCI8?
Hi, how do i threat ORACLE exceptions in a external langue, such as Java or PHP or C

Snagnever: The same way you'd treat any exception in that language.

Xgc i can't see how that on php's oci8 api
if a exception is raised by oracle
it simply shows a warning saying that an user exception wasn't caught
but php doesnt throw an exception
and oracle doesnt say WHICH exception

Snagnever: It's best to show the exact code that caused the exception and the exact message generated.

Xgc, look
i have a package
and it have its own exceptions.
when calling it inside php, if it raises an exception, oracle says:

Snagnever: Because it sounds more like you have an oracle exception that was not handled... and your API doesn't pass the exception to the php layer.

yes! that's the problem, the api is not passing it!
wasn't it supposed to pass?
do i have to set an error code with pragma exception_init and treat them by myself?

did you even bother looking at ocierror() ?

Snagnever: or you are just not catching it at the php layer.

RJarett

'code' and 'message' within it

it's always ORA-06510
'user exception not handled'
if i use the exception_init trick
the error code is the same

is your php calling plsql?

but on the error message, it says 'exception -90001 not handled bla bla bla'
RJarett yes
it's calling a package procedure

what do you get if you call the package/proc in sqlplus?

same thing
same error, same message

so it is in no way related to php

but how to pass the exception?
how to know the actual error?