cx-oracle-users

[cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Alex T. <al...@fi...>]

Hello,

This is a question about the C implementation of cx_Oracle. I've been
reading the source code and trying to understand how you are using OCI.

When a Connection is acquired from a SessionPool, it creates a brand new
environment for itself, rather than using the environment that the
SessionPool already has. That seems odd to me. (See Connection_Acquire()
and Connection_Init() in Connection.c.)

Can you explain why you create a new environment?

Are you following the OCI documentation (if so where - I can't find
anything that suggests this) or have you established the method's
reliability by experience?

(Sorry in advance if this is a stupid question.)

-Alex

-- 
:: Let me solve your problems: http://www.firetree.net/consulting/
:: alex.tingle AT firetree.net  +44-7901-552763 

Re: [cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Anthony T. <ant...@gm...>]

It is not a stupid question. I'm not absolutely certain of the answer,
either. :-)

If I understand correctly, however, the environment handle is the
"master" handle and all of the other handles are children to it. In
order to establish thread safety you want to make sure that the
handles from one thread are not used in another -- otherwise you need
to establish locking and the like which has its own set of issues.
Does that answer your question? May I ask why you asked the question
in the first place?

On 9/18/06, Alex Tingle <al...@fi...> wrote:
> Hello,
>
> This is a question about the C implementation of cx_Oracle. I've been
> reading the source code and trying to understand how you are using OCI.
>
> When a Connection is acquired from a SessionPool, it creates a brand new
> environment for itself, rather than using the environment that the
> SessionPool already has. That seems odd to me. (See Connection_Acquire()
> and Connection_Init() in Connection.c.)
>
> Can you explain why you create a new environment?
>
> Are you following the OCI documentation (if so where - I can't find
> anything that suggests this) or have you established the method's
> reliability by experience?
>
> (Sorry in advance if this is a stupid question.)
>
> -Alex
>
> --
> :: Let me solve your problems: http://www.firetree.net/consulting/
> :: alex.tingle AT firetree.net  +44-7901-552763
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>

Re: [cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Alex T. <al...@fi...>]

Hi Anthony,

On 19 Sep 2006, at 14:49, Anthony Tuininga wrote:

> May I ask why you asked the question in the first place?

I'll answer the last question first: I'm toying with the idea of  
'porting' cx_Oracle to be a C++ interface to OCI. It's really and  
exercise in teaching myself to use OCI. The practical result is I'm  
reading the source code, and trying to work out what it's doing.

The code I'm asking about puzzled me. It was either very subtle, or a  
potential bug. If it's subtle, I wanted to know. If it's a bug, I  
thought you'd want to know.


> If I understand correctly, however, the environment handle is the
> "master" handle and all of the other handles are children to it.

That seems reasonable. But if the environment handle is the "master" -  
how can it be right to make a new environment for a session handle  
that's been acquired from a pool that already HAS an environment?


> In order to establish thread safety you want to make sure that the
> handles from one thread are not used in another -- otherwise you need
> to establish locking and the like which has its own set of issues.

cx_Oracle's Environment struct contains an errorHandle as well as an  
env handle. Each thread certainly needs its own error handle. Perhaps a  
new Environment struct was made in order to make a new error handle,  
and the new env handle was just made by mistake?

I assume that it works OK, even if it's strictly incorrect.

If you think that cx_Oracle is wrong, then I might send you a patch, if  
you like.

-Alex

--

> On 9/18/06, Alex Tingle <al...@fi...> wrote:
>> Hello,
>>
>> This is a question about the C implementation of cx_Oracle. I've been
>> reading the source code and trying to understand how you are using  
>> OCI.
>>
>> When a Connection is acquired from a SessionPool, it creates a brand  
>> new
>> environment for itself, rather than using the environment that the
>> SessionPool already has. That seems odd to me. (See  
>> Connection_Acquire()
>> and Connection_Init() in Connection.c.)
>>
>> Can you explain why you create a new environment?
>>
>> Are you following the OCI documentation (if so where - I can't find
>> anything that suggests this) or have you established the method's
>> reliability by experience?
>>
>> (Sorry in advance if this is a stupid question.)
>>
>> -Alex
>>
>> --
>> :: Let me solve your problems: http://www.firetree.net/consulting/
>> :: alex.tingle AT firetree.net  +44-7901-552763
>>
>> ---------------------------------------------------------------------- 
>> ---
>> Take Surveys. Earn Cash. Influence the Future of IT
>> Join SourceForge.net's Techsay panel and you'll get the chance to  
>> share your
>> opinions on IT & business topics through brief surveys -- and earn  
>> cash
>> http://www.techsay.com/default.php? 
>> page=join.php&p=sourceforge&CID=DEVDEV
>> _______________________________________________
>> cx-oracle-users mailing list
>> cx-...@li...
>> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>>
>
> ----------------------------------------------------------------------- 
> --
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to  
> share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php? 
> page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users

Re: [cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Anthony T. <ant...@gm...>]

On 9/19/06, Alex Tingle <al...@fi...> wrote:
> Hi Anthony,
>
> On 19 Sep 2006, at 14:49, Anthony Tuininga wrote:
>
> > May I ask why you asked the question in the first place?
>
> I'll answer the last question first: I'm toying with the idea of
> 'porting' cx_Oracle to be a C++ interface to OCI. It's really and
> exercise in teaching myself to use OCI. The practical result is I'm
> reading the source code, and trying to work out what it's doing.

Ok. Just be aware that the OCI is a jungle and you need a good sharp
machete to make your way through it... :-) The part that is the most
convoluted is the whole binding/defining area -- the rules are very
obscure and often found only by trial and error. I'm very happy to not
have to deal with the OCI very much at all on a regular basis!

> The code I'm asking about puzzled me. It was either very subtle, or a
> potential bug. If it's subtle, I wanted to know. If it's a bug, I
> thought you'd want to know.

Sure. If its a bug I'd certainly like to know about it. I don't
believe it is, though. We have an application that is heavily threaded
and uses session pools and it appears to be very stable. We've run it
overnight with about 10-15 threads constantly making queries and it
has worked flawlessly.

> > If I understand correctly, however, the environment handle is the
> > "master" handle and all of the other handles are children to it.
>
> That seems reasonable. But if the environment handle is the "master" -
> how can it be right to make a new environment for a session handle
> that's been acquired from a pool that already HAS an environment?

Not certain. Oracle doesn't provide much information about these sorts
of things in the OCI. Its one of those things you simply try -- and if
it works you're happy. I can't remember if I tried the other method
but I think I did and it didn't work very well. Feel free to try it
yourself, though, as my memory may be faulty. :-)

> > In order to establish thread safety you want to make sure that the
> > handles from one thread are not used in another -- otherwise you need
> > to establish locking and the like which has its own set of issues.
>
> cx_Oracle's Environment struct contains an errorHandle as well as an
> env handle. Each thread certainly needs its own error handle. Perhaps a
> new Environment struct was made in order to make a new error handle,
> and the new env handle was just made by mistake?

Maybe, but I don't think so.

> I assume that it works OK, even if it's strictly incorrect.

It does work OK. I'm not sure if it is strictly incorrect or not. If
you can find a pointer in the OCI documentation about this I'd be
happy to see it.

> If you think that cx_Oracle is wrong, then I might send you a patch, if
> you like.

Try it first and let me know how it works for you. :-)

> -Alex
>
> --
>
> > On 9/18/06, Alex Tingle <al...@fi...> wrote:
> >> Hello,
> >>
> >> This is a question about the C implementation of cx_Oracle. I've been
> >> reading the source code and trying to understand how you are using
> >> OCI.
> >>
> >> When a Connection is acquired from a SessionPool, it creates a brand
> >> new
> >> environment for itself, rather than using the environment that the
> >> SessionPool already has. That seems odd to me. (See
> >> Connection_Acquire()
> >> and Connection_Init() in Connection.c.)
> >>
> >> Can you explain why you create a new environment?
> >>
> >> Are you following the OCI documentation (if so where - I can't find
> >> anything that suggests this) or have you established the method's
> >> reliability by experience?
> >>
> >> (Sorry in advance if this is a stupid question.)
> >>
> >> -Alex
> >>
> >> --
> >> :: Let me solve your problems: http://www.firetree.net/consulting/
> >> :: alex.tingle AT firetree.net  +44-7901-552763
> >>
> >> ----------------------------------------------------------------------
> >> ---
> >> Take Surveys. Earn Cash. Influence the Future of IT
> >> Join SourceForge.net's Techsay panel and you'll get the chance to
> >> share your
> >> opinions on IT & business topics through brief surveys -- and earn
> >> cash
> >> http://www.techsay.com/default.php?
> >> page=join.php&p=sourceforge&CID=DEVDEV
> >> _______________________________________________
> >> cx-oracle-users mailing list
> >> cx-...@li...
> >> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
> >>
> >
> > -----------------------------------------------------------------------
> > --
> > Take Surveys. Earn Cash. Influence the Future of IT
> > Join SourceForge.net's Techsay panel and you'll get the chance to
> > share your
> > opinions on IT & business topics through brief surveys -- and earn cash
> > http://www.techsay.com/default.php?
> > page=join.php&p=sourceforge&CID=DEVDEV
> > _______________________________________________
> > cx-oracle-users mailing list
> > cx-...@li...
> > https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>

Re: [cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Alex T. <al...@fi...>]

Hi Anthony,

Thanks for taking the time to respond.

On 19 Sep 2006, at 15:31, Anthony Tuininga wrote:

>> If you think that cx_Oracle is wrong, then I might send you a patch, 
>> if
>> you like.
>
> Try it first and let me know how it works for you. :-)

Of course. That's always the first and last rule ;-)

Is there a test suite?

-Alex

--

Re: [cx-oracle-users] Why do Connection and SessionPool objects not share an Environment?

[Anthony T. <ant...@gm...>]

On 9/19/06, Alex Tingle <al...@fi...> wrote:
> Hi Anthony,
>
> Thanks for taking the time to respond.

You're welcome.

> On 19 Sep 2006, at 15:31, Anthony Tuininga wrote:
>
> >> If you think that cx_Oracle is wrong, then I might send you a patch,
> >> if
> >> you like.
> >
> > Try it first and let me know how it works for you. :-)
>
> Of course. That's always the first and last rule ;-)
>
> Is there a test suite?

Yes. Its in the "test" subdirectory. That covers the basics. It does
not cover (of course) running something with many threads for a long
period of time. That you'll have to set up yourself.

> -Alex
>
> --
>
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>