Looks good - I'm running Minerva w/Oracle and I can see that I'm reusing
preparedStatements.
Cheers
-Original Message-
From: Rickard Öberg [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 30, 2000 10:48 AM
To: jBoss
Subject: Re: [jBoss-User] ejb-ref question
Aaron Mu
Aaron Mulder wrote:
>
> Ahhh I just misunderstood your point entirely. Sure, we
> could have every Connection cache it's PreparedStatements. Again, I don't
> think that will be a major gain, but it'll take about 2 seconds so I can
> go ahead and put it in.
Major coolness Aaron ;-)
Ahhh I just misunderstood your point entirely. Sure, we
could have every Connection cache it's PreparedStatements. Again, I don't
think that will be a major gain, but it'll take about 2 seconds so I can
go ahead and put it in.
Aaron
On Wed, 30 Aug 2000, Rickard [iso-8859-1] Öberg
Mulder [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 30, 2000 9:19 AM
To: jBoss
Subject: Re: [jBoss-User] ejb-ref question
This message uses a character set that is not supported by the Internet
Service. To view the original message content, open the attached message.
If the text doesn't di
Hey
Aaron Mulder wrote:
> On Thu, 24 Aug 2000, Rickard [iso-8859-1] Öberg wrote:
> > ...
> > One thing that needs to be done is pooling of prepared statements. That
> > should easily be added to Minerva I think. Aaron, any thoughts?
>
> Sure but... Minerva treats everything in the pool
On Thu, 24 Aug 2000, Rickard [iso-8859-1] Öberg wrote:
> ...
> One thing that needs to be done is pooling of prepared statements. That
> should easily be added to Minerva I think. Aaron, any thoughts?
Sure but... Minerva treats everything in the pool as identical,
so you would have to s
Hey
Athlon 650MHz/256RAM.
marc fleury wrote:
> Ok testing on PIII500 256meg RAM, it's a machine with no name ...
>
> with HotSpot Client (win2000):
>
> stateless: 4.4ms
2ms
> stateless with jndi lookup from bean: 6.1ms (!:)))
>
> stateful: 8.2ms
>
> entity CMP: 7.1ms (!!) (store a string)
Hey
Sebastien Alborini wrote:
> I don't agree with your fix:
>
> - bind(ctx, ref.getName(), new LinkRef(ref.getLink()));
> + bind(ctx, ref.getName(), new
>
>LinkRef(getApplication().getContainer(ref.getLink()).getBeanMetaData().getJndiName()));
>
> For this to work, the referenced bean must al
Hey
marc fleury wrote:
> > marc fleury wrote:
> > > Seriously now. I believe we still have a serious problem with
> > performance.
> > > I need to double check that and make sure those numbers are
> > real. But 10ms
> > > for Stateless beans is not really good, in fact it is pretty bad...
> >
>
tres bien conjugue :)
Love R2D2!
marc
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Sebastien Alborini
> Sent: Wednesday, August 23, 2000 2:33 PM
> To: jBoss
> Subject: Re: [jBoss-User] ejb-ref question
>
>
>
Sebastien Alborini wrote:
>
> (btw this is broken anyway for compound names, we should create
> intermediate subcontexts.)
Forget that. I just saw the code.
Sebastien
--
--
To subscribe:[EMAIL PROTECTED]
To unsubscribe:
d, we are
> > beta...
> > marc
>
> > > -Original Message-
> > > From: [EMAIL PROTECTED]
> > > [mailto:[EMAIL PROTECTED]]On Behalf Of marc fleury
> > > Sent: Wednesday, August 23, 2000 11:43 AM
> > > To: jBoss
> >
Hi!
Rickard Öberg wrote:
> Ok, "fixed". I actually did what I told Sebastien not to do: use
> LinkRefs to the global JNDI namespace. Works, but not perfect since it
> assumes globally bound JNDI beans (ok, a pedantic thing, but still).
I don't agree with your fix:
- bind(ctx, ref.getName(), n
;s actually decent and pretty good, we are
> beta...
> marc
> > -Original Message-
> > From: [EMAIL PROTECTED]
> > [mailto:[EMAIL PROTECTED]]On Behalf Of marc fleury
> > Sent: Wednesday, August 23, 2000 11:43 AM
> > To: jBoss
> > Subject: RE: [jBoss-Us
y, August 23, 2000 11:43 AM
> To: jBoss
> Subject: RE: [jBoss-User] ejb-ref question
>
>
>
>
> > -Original Message-
> > From: [EMAIL PROTECTED]
> > [mailto:[EMAIL PROTECTED]]On Behalf Of Rickard Öberg
> > Sent: Wednesday, August 23, 2000 11:20 AM
&g
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Rickard Öberg
> Sent: Wednesday, August 23, 2000 11:20 AM
> To: jBoss
> Subject: Re: [jBoss-User] ejb-ref question
>
>
> Hey
>
> marc fleury wrote:
> > Se
ad us, we're happy :)))
dude,
marc
PS: yeah... come help... :)
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Jeff McArthur
> Sent: Wednesday, August 23, 2000 10:37 AM
> To: 'jBoss'
> Subject: RE: [jBoss-User] e
Hey
marc fleury wrote:
> Seriously now. I believe we still have a serious problem with performance.
> I need to double check that and make sure those numbers are real. But 10ms
> for Stateless beans is not really good, in fact it is pretty bad...
That's weird, I get 2-3 ms per call..
/R
--
> Your design is real rickard, it is a success and the integration
> proves it.
Oh and in case I haven't said it yet,
I truly believe your design is better
If only you would let us help you make it work...
you would be amazed, ease of use and all, wow!...
kind regards
marc
>
> I am also sligh
p.s. - Dude, not everyone from California talks like that, dude, unless it's
after the _free_ beer mentioned in the FAQ! Dudes, free beer!
From: Rickard Öberg [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 23, 2000 8:40 AM
To: jBoss
Subject: Re: [jBoss-User] ejb-ref question
m
> I have no idea what you are talking about. Oh, but that's true. Back in
> EJBoss 1.0 I do remember the "container" being in more or less 1 class.
> So, guilty as charged then...
that's right, I am actually very proud of that fact and the fact that it
(basically the container and the wrapper)
marc fleury wrote:
> Ok! I will give you that one, one or two depths but no more!
Never said anything else, did I?
> you get me
> every time you devilish little thing I get all puffed up and I say "No
> Rickard!" (tail wagging) "No more 10 depth hierarchiesRickard!" but
> then you get
Hi!
marc fleury wrote:
> > Well, as I said it works, but it makes assumptions that I'm not really
> > comfortable with.
>
> You mean that the stuff is deployed already ?
No, that there is a JNDI-namespace. Not required...
Until someone wants to make a microcontainer that doesn't do
distributio
ly interface a newbie should implement is
implements "reliable, readable, simple" :
regards
marc
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Rickard Öberg
> Sent: Wednesday, August 23, 2000 8:40 AM
> To: jBoss
&g
ut which things should be
in use and which not.
Cheers
-Original Message-
From: Rickard Öberg [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 23, 2000 11:40 AM
To: jBoss
Subject: Re: [jBoss-User] ejb-ref question
marc fleury wrote:
> you do that in other parts of the code (f
marc fleury wrote:
> you do that in other parts of the code (factor out metadata information) and
> I don't think it is a good idea. Byproduct of "design talk" but not real.
> So if the variable is in the metadata and you don't need it for
> MessageBeans? well don't use it!!! but don't put a hier
> Well, as I said it works, but it makes assumptions that I'm not really
> comfortable with.
You mean that the stuff is deployed already ?
don't worry about that that was my "ln -s virtual" point.
It is just a String referal and it "lazy loaded" afaik.
marc
>
> Anyway, until someone complains
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Ingo Bruell
> Sent: Wednesday, August 23, 2000 4:16 AM
> To: jBoss
> Subject: Re: [jBoss-User] ejb-ref question
>
>
> Hi
>
> Ok, "fixed". I actual
Hi!
marc fleury wrote:
> Glad you decided to override your own recommendation for the sake of
> "simple"... we might be getting somewhere here :)
Well, as I said it works, but it makes assumptions that I'm not really
comfortable with.
Anyway, until someone complains about it, whatever works wo
That is ok rickard, if you played with unix a bit you would know about
ln -s.
Sebastien showed me that Linux takes an ln -s even when the target file is
virtual, I did not know that... extremelly convenient for administration.
Think of it the same way, LinkRef is your ln -s.
It is less "pedantic
Hi
Ok, "fixed". I actually did what I told Sebastien not to do: use
LinkRefs to the global JNDI namespace. Works, but not perfect since it
assumes globally bound JNDI beans (ok, a pedantic thing, but still).
Also, while doing this I noticed that BeanMetaData holds way too many
variables that are
Hey
Rickard Öberg wrote:
> Ingo Bruell wrote:
> > after I have updated jBoss to the actual CVS (today morning) my SessionBeans
> > could not find other Beans referenced in the descriptor.
> > How I have to declare the references in the descriptor ?
>
> Just checked. The ejb-ref handling was chan
Hey
Ingo Bruell wrote:
> after I have updated jBoss to the actual CVS (today morning) my SessionBeans
> could not find other Beans referenced in the descriptor.
> How I have to declare the references in the descriptor ?
Just checked. The ejb-ref handling was changed yesterday, and it is
currentl
33 matches
Mail list logo