You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-dev@axis.apache.org by Aleksander Slominski <as...@cs.indiana.edu> on 2004/12/03 08:03:47 UTC
[Axis2] Re: Full infoset support
i know i may be PITA but please do put [Axis2] somewhere in email
subject ...
my reply is below.
Ajith Ranabahu wrote:
>Since this issue was taken up in the last IRC session and ended with
>no conclusion, I suppose we should talks about this seriously and take
>a decision. So I am going to put forward the qeustion in simple terms
>and put up my answer with reasons.
>
>Q - Do we need full XML infoset support with OM?
>
>A - No.
>
>
please define what is exactly full XML infoset support?
for me it is: AXIOM API must support XML Infoset and allow an API
implementation that implements all XML infoset if it finds it necessary
but default OM implementation is optimized for SOAP.
i think Web Services are more than just SOAP and precluding use of
XML/HTTP would be an error (especially that it is so easy to support).
the only parts of XML Infoset that we may not want to support are those
not required by SOAP (mostly related to PIs and DTDs as they are
prohibited in SOAP 1.2 message for details see
http://www.w3.org/TR/soap12-part1/#soapenv - anything more excluded? and
comments are allowed just not inside message but in document ...)
so think it can be simply done by having DTD related properties
empty/default/null (such as those in
http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl to
throw exception if it sees DTD (configurable to do so as required for
SOAP 1.2 ...)
this is very quick analysis - let me know if i missed something.
thanks,
alek
--
The best way to predict the future is to invent it - Alan Kay
Re: [Axis2] Re: Full infoset support
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
Eran Chinthaka wrote:
>
>
>>>i know i may be PITA but please do put [Axis2] somewhere in email
>>>subject ...
>>>
>>>
>[Chinthaka] Alek, Axis2 was there in the subject. My email filters got it
>correct.
>
>
not the email i was repsonding - i know see that it had followup ...
alek
--
The best way to predict the future is to invent it - Alan Kay
Re: [Axis2] Re: Full infoset support
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
Sanjiva Weerawarana wrote:
>"Aleksander Slominski" <as...@cs.indiana.edu> writes:
>
>
>>this is another reason why people that wants true doc/lit exchange will
>>want MTOM or look for alternative and simple XML/HTTP
>>
>>
>
>Yes.
>
>
>
>>the only safe way to transport XML document inside another XML document
>>is escaping it as string or even better BASE64 (in SOAP 1.1 there were
>>discussions about "id" attribute ...)
>>
>>
>
>Yes.
>
>But none of this argues for making the OM have stuff that is not
>needed for SOAP. Well ok, if you have an MTOM thing which is a full
>XML document I guess it would be nice to be able to use the OM
>API for that XML document too.
>
yes
>We should consider that carefully
>when designing the MTOM support for OM.
>
>
yes!
alek
--
The best way to predict the future is to invent it - Alan Kay
Re: [Axis2] Re: Full infoset support
Posted by Sanjiva Weerawarana <sa...@opensource.lk>.
"Aleksander Slominski" <as...@cs.indiana.edu> writes:
> this is another reason why people that wants true doc/lit exchange will
> want MTOM or look for alternative and simple XML/HTTP
Yes.
> the only safe way to transport XML document inside another XML document
> is escaping it as string or even better BASE64 (in SOAP 1.1 there were
> discussions about "id" attribute ...)
Yes.
But none of this argues for making the OM have stuff that is not
needed for SOAP. Well ok, if you have an MTOM thing which is a full
XML document I guess it would be nice to be able to use the OM
API for that XML document too. We should consider that carefully
when designing the MTOM support for OM.
Sanjiva.
Re: [Axis2] Re: Full infoset support
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
this is another reason why people that wants true doc/lit exchange will
want MTOM or look for alternative and simple XML/HTTP
the only safe way to transport XML document inside another XML document
is escaping it as string or even better BASE64 (in SOAP 1.1 there were
discussions about "id" attribute ...)
alek
Davanum Srinivas wrote:
>ok understood.
>
>
>On Fri, 3 Dec 2004 22:08:03 +0600, Sanjiva Weerawarana
><sa...@opensource.lk> wrote:
>
>
>>A PI ..
>>
>> <Body><? blah .. ?> </Body>
>>
>>is not legal SOAP even tho its valid XML.
>>
>>
>>
>>Sanjiva.
>>
>>----- Original Message -----
>>From: "Davanum Srinivas" <da...@gmail.com>
>>To: <ax...@ws.apache.org>
>>Cc: "Ajith Ranabahu" <aj...@gmail.com>
>>Sent: Friday, December 03, 2004 7:40 PM
>>Subject: Re: [Axis2] Re: Full infoset support
>>
>>
>>
>>>Hmmm. my bad then...Can you please give me an example of what SOAP
>>>does not allow inside <Body>...</Body>? (but something which is part
>>>of the "full XML Infoset"?)
>>>
>>>thanks,
>>>dims
>>>
>>>
>>>
>>>On Fri, 3 Dec 2004 19:27:34 +0600, Sanjiva Weerawarana
>>><sa...@opensource.lk> wrote:
>>>
>>>
>>>>That's not true Dims .. SOAP doesn't *allow* you to transmit the
>>>>full XML Infoset. So if you're doing SAAJ then you will *not*
>>>>have the full Infoset present, even though the DOM behind SAAJ
>>>>is capable of representing the full Infoset.
>>>>
>>>>The OM as it stands *is* capable of representing *any* XML that
>>>>can appear inside <Body>...</Body>.
>>>>
>>>>Sanjiva.
>>>>
>>>>
>>>>
>>>>----- Original Message -----
>>>>From: "Davanum Srinivas" <da...@gmail.com>
>>>>To: <ax...@ws.apache.org>; "Ajith Ranabahu"
>>>>
>>>>
>><aj...@gmail.com>
>>
>>
>>>>Sent: Friday, December 03, 2004 6:49 PM
>>>>Subject: Re: [Axis2] Re: Full infoset support
>>>>
>>>>
>>>>
>>>>>Ajith,
>>>>>
>>>>>Please remember, if the user wants to do doc/lit by hand....It implies
>>>>>he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
>>>>>So If you can't do complete XML Infoset there....then OM is useless
>>>>>for him. Also remember you can't do SAAJ if OM does not support
>>>>>complete XML Infoset.
>>>>>
>>>>>-- dims
>>>>>
>>>>>
>>>>>On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
>>>>><aj...@gmail.com> wrote:
>>>>>
>>>>>
>>>>>>Hi there Alek,
>>>>>>Sorry about the first mail which actually lacked the "Axis2" part.
>>>>>>
>>>>>>
>>It
>>
>>
>>>>>>was sent in by mistake.
>>>>>>BTW i think the rationale behind making OM specific should be
>>>>>>mentioned in a bit more detail here. This is how I see it so if
>>>>>>
>>>>>>
>>other
>>
>>
>>>>>>guys feel I am wrong please correct me.
>>>>>>Ok here goes
>>>>>>1. XML infoset is truly fundamental. No doubt about it. However that
>>>>>>fundamentality only plays a role if the tool we are making is pure
>>>>>>
>>>>>>
>>XML
>>
>>
>>>>>>based. In this case our tool (Axis 2 to be specific) is focused into
>>>>>>
>>>>>>
>>a
>>
>>
>>>>>>specific subset of XML which is SOAP. So my point is that what we
>>>>>>should be making is a SOAP object model (that is what OM is right
>>>>>>now). You can just call it a bear-bones SOAP object model.
>>>>>>2.We want this OM to be light weight. That is why we refrained from
>>>>>>the easy choice, making OM on top of something like XOM or JDOM. We
>>>>>>can incorporate the differed building capability to such object
>>>>>>
>>>>>>
>>models
>>
>>
>>>>>>without much hassel. However that wont solve our memory footprint
>>>>>>problem. Hence OM was written purely from scratch to include the
>>>>>>eseential information items and minimize the content in the memory.
>>>>>>3. Well we DO gain a lot, in terms of memory,simplicity and speed by
>>>>>>catering for SOAP only. The concern of many people is not another
>>>>>>complete XML infoset representation (they already have plenty of
>>>>>>choices like DOM,JDOM,XOM or even XML beans which has complete
>>>>>>
>>>>>>
>>schema
>>
>>
>>>>>>support). However there are very few SOAP stacks!
>>>>>>
>>>>>>Hope this helps to get our rationale through :)
>>>>>>
>>>>>>On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
>>>>>>
>>>>>>
>>>>>><as...@cs.indiana.edu> wrote:
>>>>>>
>>>>>>
>>>>>>>Eran Chinthaka wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>>>for me it is: AXIOM API must support XML Infoset and allow an
>>>>>>>>>>
>>>>>>>>>>
>>API
>>
>>
>>>>>>>>>>implementation that implements all XML infoset if it finds it
>>>>>>>>>>
>>>>>>>>>>
>>>>necessary
>>>>
>>>>
>>>>>>>>>>but default OM implementation is optimized for SOAP.
>>>>>>>>>>
>>>>>>>>>>i think Web Services are more than just SOAP and precluding use
>>>>>>>>>>
>>>>>>>>>>
>>of
>>
>>
>>>>>>>>>>XML/HTTP would be an error (especially that it is so easy to
>>>>>>>>>>
>>>>>>>>>>
>>>>support).
>>>>
>>>>
>>>>>>>>>>the only parts of XML Infoset that we may not want to support
>>>>>>>>>>
>>>>>>>>>>
>>are
>>
>>
>>>>those
>>>>
>>>>
>>>>>>>>>>not required by SOAP (mostly related to PIs and DTDs as they
>>>>>>>>>>
>>>>>>>>>>
>>are
>>
>>
>>>>>>>>>>prohibited in SOAP 1.2 message for details see
>>>>>>>>>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
>>>>>>>>>>
>>>>>>>>>>
>>>>excluded? and
>>>>
>>>>
>>>>>>>>>>comments are allowed just not inside message but in document
>>>>>>>>>>
>>>>>>>>>>
>>...)
>>
>>
>>>>>>>>>>so think it can be simply done by having DTD related properties
>>>>>>>>>>empty/default/null (such as those in
>>>>>>>>>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM
>>>>>>>>>>
>>>>>>>>>>
>>impl
>>
>>
>>>>to
>>>>
>>>>
>>>>>>>>>>throw exception if it sees DTD (configurable to do so as
>>>>>>>>>>
>>>>>>>>>>
>>required
>>
>>
>>>>for
>>>>
>>>>
>>>>>>>>>>SOAP 1.2 ...)
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>[Chinthaka] can u all remember the basic concepts we agreed on
>>>>>>>>
>>>>>>>>
>>the
>>
>>
>>>>summit.
>>>>
>>>>
>>>>>>>>One is YAGNI and one other is KISS. So if we just create some
>>>>>>>>
>>>>>>>>
>>support
>>
>>
>>>>for
>>>>
>>>>
>>>>>>>>DTD and PI and just throw an exception, isn't it YAGNI or
>>>>>>>>
>>>>>>>>
>>violating
>>
>>
>>>>KISS
>>>>
>>>>
>>>>>>>>???????
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>i regard XML infoset API as _fundamental_ (so it is YAGNI) and
>>>>>>>
>>>>>>>
>>not
>>
>>
>>>>>>>something that can be simplified, omitted, or further KISSed.
>>>>>>>
>>>>>>>this is different from actual OM impls that may chose to not
>>>>>>>
>>>>>>>
>>support
>>
>>
>>>>>>>something in XML Infoset and throw exception (like in case of
>>>>>>>
>>>>>>>
>>SOAP)
>>
>>
>>>>>>>therefore i do not think we differ much - i think XML Infoset API
>>>>>>>
>>>>>>>
>>such
>>
>>
>>>>>>>as AXIOM should be reflecting XML Infoset element information
>>>>>>>
>>>>>>>
>>items
>>
>>
>>>>and
>>>>
>>>>
>>>>>>>properties but implementations can be optimized for subset of XML
>>>>>>>Infoset (such as SOAP 1.2 or 1.1).
>>>>>>>
>>>>>>>however i do not see any considerable gains by fusing SOAP and XML
>>>>>>>Infoset APIs together but it will make things bigger and more
>>>>>>>complicated ...
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>alek
>>>>>>>
>>>>>>>--
>>>>>>>The best way to predict the future is to invent it - Alan Kay
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>--
>>>>>>Ajith Ranabahu
>>>>>>
>>>>>>
>>>>>>
>>>>>--
>>>>>Davanum Srinivas - http://webservices.apache.org/~dims/
>>>>>
>>>>>
>>>>
>>>>
>>>--
>>>Davanum Srinivas - http://webservices.apache.org/~dims/
>>>
>>>
>>
>>
>
>
>
>
--
The best way to predict the future is to invent it - Alan Kay
Re: [Axis2] Re: Full infoset support
Posted by Davanum Srinivas <da...@gmail.com>.
ok understood.
On Fri, 3 Dec 2004 22:08:03 +0600, Sanjiva Weerawarana
<sa...@opensource.lk> wrote:
> A PI ..
>
> <Body><? blah .. ?> </Body>
>
> is not legal SOAP even tho its valid XML.
>
>
>
> Sanjiva.
>
> ----- Original Message -----
> From: "Davanum Srinivas" <da...@gmail.com>
> To: <ax...@ws.apache.org>
> Cc: "Ajith Ranabahu" <aj...@gmail.com>
> Sent: Friday, December 03, 2004 7:40 PM
> Subject: Re: [Axis2] Re: Full infoset support
>
> > Hmmm. my bad then...Can you please give me an example of what SOAP
> > does not allow inside <Body>...</Body>? (but something which is part
> > of the "full XML Infoset"?)
> >
> > thanks,
> > dims
> >
> >
> >
> > On Fri, 3 Dec 2004 19:27:34 +0600, Sanjiva Weerawarana
> > <sa...@opensource.lk> wrote:
> > > That's not true Dims .. SOAP doesn't *allow* you to transmit the
> > > full XML Infoset. So if you're doing SAAJ then you will *not*
> > > have the full Infoset present, even though the DOM behind SAAJ
> > > is capable of representing the full Infoset.
> > >
> > > The OM as it stands *is* capable of representing *any* XML that
> > > can appear inside <Body>...</Body>.
> > >
> > > Sanjiva.
> > >
> > >
> > >
> > > ----- Original Message -----
> > > From: "Davanum Srinivas" <da...@gmail.com>
> > > To: <ax...@ws.apache.org>; "Ajith Ranabahu"
> <aj...@gmail.com>
> > > Sent: Friday, December 03, 2004 6:49 PM
> > > Subject: Re: [Axis2] Re: Full infoset support
> > >
> > > > Ajith,
> > > >
> > > > Please remember, if the user wants to do doc/lit by hand....It implies
> > > > he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
> > > > So If you can't do complete XML Infoset there....then OM is useless
> > > > for him. Also remember you can't do SAAJ if OM does not support
> > > > complete XML Infoset.
> > > >
> > > > -- dims
> > > >
> > > >
> > > > On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
> > > > <aj...@gmail.com> wrote:
> > > > > Hi there Alek,
> > > > > Sorry about the first mail which actually lacked the "Axis2" part.
> It
> > > > > was sent in by mistake.
> > > > > BTW i think the rationale behind making OM specific should be
> > > > > mentioned in a bit more detail here. This is how I see it so if
> other
> > > > > guys feel I am wrong please correct me.
> > > > > Ok here goes
> > > > > 1. XML infoset is truly fundamental. No doubt about it. However that
> > > > > fundamentality only plays a role if the tool we are making is pure
> XML
> > > > > based. In this case our tool (Axis 2 to be specific) is focused into
> a
> > > > > specific subset of XML which is SOAP. So my point is that what we
> > > > > should be making is a SOAP object model (that is what OM is right
> > > > > now). You can just call it a bear-bones SOAP object model.
> > > > > 2.We want this OM to be light weight. That is why we refrained from
> > > > > the easy choice, making OM on top of something like XOM or JDOM. We
> > > > > can incorporate the differed building capability to such object
> models
> > > > > without much hassel. However that wont solve our memory footprint
> > > > > problem. Hence OM was written purely from scratch to include the
> > > > > eseential information items and minimize the content in the memory.
> > > > > 3. Well we DO gain a lot, in terms of memory,simplicity and speed by
> > > > > catering for SOAP only. The concern of many people is not another
> > > > > complete XML infoset representation (they already have plenty of
> > > > > choices like DOM,JDOM,XOM or even XML beans which has complete
> schema
> > > > > support). However there are very few SOAP stacks!
> > > > >
> > > > > Hope this helps to get our rationale through :)
> > > > >
> > > > > On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
> > > > >
> > > > >
> > > > > <as...@cs.indiana.edu> wrote:
> > > > > > Eran Chinthaka wrote:
> > > > > >
> > > > > >
> > > > > >
> > > > > > >>>for me it is: AXIOM API must support XML Infoset and allow an
> API
> > > > > > >>>implementation that implements all XML infoset if it finds it
> > > necessary
> > > > > > >>>but default OM implementation is optimized for SOAP.
> > > > > > >>>
> > > > > > >>>i think Web Services are more than just SOAP and precluding use
> of
> > > > > > >>>XML/HTTP would be an error (especially that it is so easy to
> > > support).
> > > > > > >>>
> > > > > > >>>the only parts of XML Infoset that we may not want to support
> are
> > > those
> > > > > > >>>not required by SOAP (mostly related to PIs and DTDs as they
> are
> > > > > > >>>prohibited in SOAP 1.2 message for details see
> > > > > > >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
> > > excluded? and
> > > > > > >>>comments are allowed just not inside message but in document
> ...)
> > > > > > >>>
> > > > > > >>>so think it can be simply done by having DTD related properties
> > > > > > >>>empty/default/null (such as those in
> > > > > > >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM
> impl
> > > to
> > > > > > >>>throw exception if it sees DTD (configurable to do so as
> required
> > > for
> > > > > > >>>SOAP 1.2 ...)
> > > > > > >>>
> > > > > > >>>
> > > > > > >
> > > > > > >[Chinthaka] can u all remember the basic concepts we agreed on
> the
> > > summit.
> > > > > > >One is YAGNI and one other is KISS. So if we just create some
> support
> > > for
> > > > > > >DTD and PI and just throw an exception, isn't it YAGNI or
> violating
> > > KISS
> > > > > > >???????
> > > > > > >
> > > > > > >
> > > > > > i regard XML infoset API as _fundamental_ (so it is YAGNI) and
> not
> > > > > > something that can be simplified, omitted, or further KISSed.
> > > > > >
> > > > > > this is different from actual OM impls that may chose to not
> support
> > > > > > something in XML Infoset and throw exception (like in case of
> SOAP)
> > > > > >
> > > > > > therefore i do not think we differ much - i think XML Infoset API
> such
> > > > > > as AXIOM should be reflecting XML Infoset element information
> items
> > > and
> > > > > > properties but implementations can be optimized for subset of XML
> > > > > > Infoset (such as SOAP 1.2 or 1.1).
> > > > > >
> > > > > > however i do not see any considerable gains by fusing SOAP and XML
> > > > > > Infoset APIs together but it will make things bigger and more
> > > > > > complicated ...
> > > > > >
> > > > > >
> > > > > >
> > > > > > alek
> > > > > >
> > > > > > --
> > > > > > The best way to predict the future is to invent it - Alan Kay
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Ajith Ranabahu
> > > > >
> > > >
> > > >
> > > > --
> > > > Davanum Srinivas - http://webservices.apache.org/~dims/
> > >
> > >
> >
> >
> > --
> > Davanum Srinivas - http://webservices.apache.org/~dims/
>
>
--
Davanum Srinivas - http://webservices.apache.org/~dims/
Re: [Axis2] Re: Full infoset support
Posted by Sanjiva Weerawarana <sa...@opensource.lk>.
A PI ..
<Body><? blah .. ?> </Body>
is not legal SOAP even tho its valid XML.
Sanjiva.
----- Original Message -----
From: "Davanum Srinivas" <da...@gmail.com>
To: <ax...@ws.apache.org>
Cc: "Ajith Ranabahu" <aj...@gmail.com>
Sent: Friday, December 03, 2004 7:40 PM
Subject: Re: [Axis2] Re: Full infoset support
> Hmmm. my bad then...Can you please give me an example of what SOAP
> does not allow inside <Body>...</Body>? (but something which is part
> of the "full XML Infoset"?)
>
> thanks,
> dims
>
>
>
> On Fri, 3 Dec 2004 19:27:34 +0600, Sanjiva Weerawarana
> <sa...@opensource.lk> wrote:
> > That's not true Dims .. SOAP doesn't *allow* you to transmit the
> > full XML Infoset. So if you're doing SAAJ then you will *not*
> > have the full Infoset present, even though the DOM behind SAAJ
> > is capable of representing the full Infoset.
> >
> > The OM as it stands *is* capable of representing *any* XML that
> > can appear inside <Body>...</Body>.
> >
> > Sanjiva.
> >
> >
> >
> > ----- Original Message -----
> > From: "Davanum Srinivas" <da...@gmail.com>
> > To: <ax...@ws.apache.org>; "Ajith Ranabahu"
<aj...@gmail.com>
> > Sent: Friday, December 03, 2004 6:49 PM
> > Subject: Re: [Axis2] Re: Full infoset support
> >
> > > Ajith,
> > >
> > > Please remember, if the user wants to do doc/lit by hand....It implies
> > > he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
> > > So If you can't do complete XML Infoset there....then OM is useless
> > > for him. Also remember you can't do SAAJ if OM does not support
> > > complete XML Infoset.
> > >
> > > -- dims
> > >
> > >
> > > On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
> > > <aj...@gmail.com> wrote:
> > > > Hi there Alek,
> > > > Sorry about the first mail which actually lacked the "Axis2" part.
It
> > > > was sent in by mistake.
> > > > BTW i think the rationale behind making OM specific should be
> > > > mentioned in a bit more detail here. This is how I see it so if
other
> > > > guys feel I am wrong please correct me.
> > > > Ok here goes
> > > > 1. XML infoset is truly fundamental. No doubt about it. However that
> > > > fundamentality only plays a role if the tool we are making is pure
XML
> > > > based. In this case our tool (Axis 2 to be specific) is focused into
a
> > > > specific subset of XML which is SOAP. So my point is that what we
> > > > should be making is a SOAP object model (that is what OM is right
> > > > now). You can just call it a bear-bones SOAP object model.
> > > > 2.We want this OM to be light weight. That is why we refrained from
> > > > the easy choice, making OM on top of something like XOM or JDOM. We
> > > > can incorporate the differed building capability to such object
models
> > > > without much hassel. However that wont solve our memory footprint
> > > > problem. Hence OM was written purely from scratch to include the
> > > > eseential information items and minimize the content in the memory.
> > > > 3. Well we DO gain a lot, in terms of memory,simplicity and speed by
> > > > catering for SOAP only. The concern of many people is not another
> > > > complete XML infoset representation (they already have plenty of
> > > > choices like DOM,JDOM,XOM or even XML beans which has complete
schema
> > > > support). However there are very few SOAP stacks!
> > > >
> > > > Hope this helps to get our rationale through :)
> > > >
> > > > On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
> > > >
> > > >
> > > > <as...@cs.indiana.edu> wrote:
> > > > > Eran Chinthaka wrote:
> > > > >
> > > > >
> > > > >
> > > > > >>>for me it is: AXIOM API must support XML Infoset and allow an
API
> > > > > >>>implementation that implements all XML infoset if it finds it
> > necessary
> > > > > >>>but default OM implementation is optimized for SOAP.
> > > > > >>>
> > > > > >>>i think Web Services are more than just SOAP and precluding use
of
> > > > > >>>XML/HTTP would be an error (especially that it is so easy to
> > support).
> > > > > >>>
> > > > > >>>the only parts of XML Infoset that we may not want to support
are
> > those
> > > > > >>>not required by SOAP (mostly related to PIs and DTDs as they
are
> > > > > >>>prohibited in SOAP 1.2 message for details see
> > > > > >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
> > excluded? and
> > > > > >>>comments are allowed just not inside message but in document
...)
> > > > > >>>
> > > > > >>>so think it can be simply done by having DTD related properties
> > > > > >>>empty/default/null (such as those in
> > > > > >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM
impl
> > to
> > > > > >>>throw exception if it sees DTD (configurable to do so as
required
> > for
> > > > > >>>SOAP 1.2 ...)
> > > > > >>>
> > > > > >>>
> > > > > >
> > > > > >[Chinthaka] can u all remember the basic concepts we agreed on
the
> > summit.
> > > > > >One is YAGNI and one other is KISS. So if we just create some
support
> > for
> > > > > >DTD and PI and just throw an exception, isn't it YAGNI or
violating
> > KISS
> > > > > >???????
> > > > > >
> > > > > >
> > > > > i regard XML infoset API as _fundamental_ (so it is YAGNI) and
not
> > > > > something that can be simplified, omitted, or further KISSed.
> > > > >
> > > > > this is different from actual OM impls that may chose to not
support
> > > > > something in XML Infoset and throw exception (like in case of
SOAP)
> > > > >
> > > > > therefore i do not think we differ much - i think XML Infoset API
such
> > > > > as AXIOM should be reflecting XML Infoset element information
items
> > and
> > > > > properties but implementations can be optimized for subset of XML
> > > > > Infoset (such as SOAP 1.2 or 1.1).
> > > > >
> > > > > however i do not see any considerable gains by fusing SOAP and XML
> > > > > Infoset APIs together but it will make things bigger and more
> > > > > complicated ...
> > > > >
> > > > >
> > > > >
> > > > > alek
> > > > >
> > > > > --
> > > > > The best way to predict the future is to invent it - Alan Kay
> > > > >
> > > > >
> > > >
> > > >
> > > > --
> > > > Ajith Ranabahu
> > > >
> > >
> > >
> > > --
> > > Davanum Srinivas - http://webservices.apache.org/~dims/
> >
> >
>
>
> --
> Davanum Srinivas - http://webservices.apache.org/~dims/
Re: [Axis2] Re: Full infoset support
Posted by Davanum Srinivas <da...@gmail.com>.
Hmmm. my bad then...Can you please give me an example of what SOAP
does not allow inside <Body>...</Body>? (but something which is part
of the "full XML Infoset"?)
thanks,
dims
On Fri, 3 Dec 2004 19:27:34 +0600, Sanjiva Weerawarana
<sa...@opensource.lk> wrote:
> That's not true Dims .. SOAP doesn't *allow* you to transmit the
> full XML Infoset. So if you're doing SAAJ then you will *not*
> have the full Infoset present, even though the DOM behind SAAJ
> is capable of representing the full Infoset.
>
> The OM as it stands *is* capable of representing *any* XML that
> can appear inside <Body>...</Body>.
>
> Sanjiva.
>
>
>
> ----- Original Message -----
> From: "Davanum Srinivas" <da...@gmail.com>
> To: <ax...@ws.apache.org>; "Ajith Ranabahu" <aj...@gmail.com>
> Sent: Friday, December 03, 2004 6:49 PM
> Subject: Re: [Axis2] Re: Full infoset support
>
> > Ajith,
> >
> > Please remember, if the user wants to do doc/lit by hand....It implies
> > he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
> > So If you can't do complete XML Infoset there....then OM is useless
> > for him. Also remember you can't do SAAJ if OM does not support
> > complete XML Infoset.
> >
> > -- dims
> >
> >
> > On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
> > <aj...@gmail.com> wrote:
> > > Hi there Alek,
> > > Sorry about the first mail which actually lacked the "Axis2" part. It
> > > was sent in by mistake.
> > > BTW i think the rationale behind making OM specific should be
> > > mentioned in a bit more detail here. This is how I see it so if other
> > > guys feel I am wrong please correct me.
> > > Ok here goes
> > > 1. XML infoset is truly fundamental. No doubt about it. However that
> > > fundamentality only plays a role if the tool we are making is pure XML
> > > based. In this case our tool (Axis 2 to be specific) is focused into a
> > > specific subset of XML which is SOAP. So my point is that what we
> > > should be making is a SOAP object model (that is what OM is right
> > > now). You can just call it a bear-bones SOAP object model.
> > > 2.We want this OM to be light weight. That is why we refrained from
> > > the easy choice, making OM on top of something like XOM or JDOM. We
> > > can incorporate the differed building capability to such object models
> > > without much hassel. However that wont solve our memory footprint
> > > problem. Hence OM was written purely from scratch to include the
> > > eseential information items and minimize the content in the memory.
> > > 3. Well we DO gain a lot, in terms of memory,simplicity and speed by
> > > catering for SOAP only. The concern of many people is not another
> > > complete XML infoset representation (they already have plenty of
> > > choices like DOM,JDOM,XOM or even XML beans which has complete schema
> > > support). However there are very few SOAP stacks!
> > >
> > > Hope this helps to get our rationale through :)
> > >
> > > On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
> > >
> > >
> > > <as...@cs.indiana.edu> wrote:
> > > > Eran Chinthaka wrote:
> > > >
> > > >
> > > >
> > > > >>>for me it is: AXIOM API must support XML Infoset and allow an API
> > > > >>>implementation that implements all XML infoset if it finds it
> necessary
> > > > >>>but default OM implementation is optimized for SOAP.
> > > > >>>
> > > > >>>i think Web Services are more than just SOAP and precluding use of
> > > > >>>XML/HTTP would be an error (especially that it is so easy to
> support).
> > > > >>>
> > > > >>>the only parts of XML Infoset that we may not want to support are
> those
> > > > >>>not required by SOAP (mostly related to PIs and DTDs as they are
> > > > >>>prohibited in SOAP 1.2 message for details see
> > > > >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
> excluded? and
> > > > >>>comments are allowed just not inside message but in document ...)
> > > > >>>
> > > > >>>so think it can be simply done by having DTD related properties
> > > > >>>empty/default/null (such as those in
> > > > >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl
> to
> > > > >>>throw exception if it sees DTD (configurable to do so as required
> for
> > > > >>>SOAP 1.2 ...)
> > > > >>>
> > > > >>>
> > > > >
> > > > >[Chinthaka] can u all remember the basic concepts we agreed on the
> summit.
> > > > >One is YAGNI and one other is KISS. So if we just create some support
> for
> > > > >DTD and PI and just throw an exception, isn't it YAGNI or violating
> KISS
> > > > >???????
> > > > >
> > > > >
> > > > i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
> > > > something that can be simplified, omitted, or further KISSed.
> > > >
> > > > this is different from actual OM impls that may chose to not support
> > > > something in XML Infoset and throw exception (like in case of SOAP)
> > > >
> > > > therefore i do not think we differ much - i think XML Infoset API such
> > > > as AXIOM should be reflecting XML Infoset element information items
> and
> > > > properties but implementations can be optimized for subset of XML
> > > > Infoset (such as SOAP 1.2 or 1.1).
> > > >
> > > > however i do not see any considerable gains by fusing SOAP and XML
> > > > Infoset APIs together but it will make things bigger and more
> > > > complicated ...
> > > >
> > > >
> > > >
> > > > alek
> > > >
> > > > --
> > > > The best way to predict the future is to invent it - Alan Kay
> > > >
> > > >
> > >
> > >
> > > --
> > > Ajith Ranabahu
> > >
> >
> >
> > --
> > Davanum Srinivas - http://webservices.apache.org/~dims/
>
>
--
Davanum Srinivas - http://webservices.apache.org/~dims/
RE: [Axis2] Re: Full infoset support
Posted by Eran Chinthaka <ch...@opensource.lk>.
Sanjiva,
I completely agree with you.
I think the current OM impl, found under svn/scratch/prototype2 area, is
covering only the things that is absolutely required for Axis 2 (except MTOM
support which will come soon)!!
-- Eran Chinthaka
>>
>>"Aleksander Slominski" <as...@cs.indiana.edu> writes:
>>> you mean API or implementation. I would like to make this distinction:
>>> API projects XML Infoset view and implementation decides what is valid
>>> in this view (for example: SOAP specific impl of XML Infoset PIs are
>>> not allowed).
>>
>>I agree with the principle of complete API + restricted impl.
>>
>>*However*, let's look at the problem we are trying to solve: We
>>are trying to build a SOAP engine. In that world, a full Infoset
>>API is YAGNI. So, while the OM can easily be extended to cover
>>the full Infoset, one has to ask the question, "Why?". Making it
>>cover the full Infoset is not a requirement to solve the problem
>>we have to solve.
>>
>>Later on, if someone were to offer a patch to "complete" the OM
>>then I'd have no problem. I do have a problem with us trying to
>>make the first snapshot of Axis2 have *anything* that is not
>>an absolutely required component of the solution.
>>
>>Sanjiva.
Re: [Axis2] Re: Full infoset support
Posted by Sanjiva Weerawarana <sa...@opensource.lk>.
"Aleksander Slominski" <as...@cs.indiana.edu> writes:
> you mean API or implementation. I would like to make this distinction:
> API projects XML Infoset view and implementation decides what is valid
> in this view (for example: SOAP specific impl of XML Infoset PIs are
> not allowed).
I agree with the principle of complete API + restricted impl.
*However*, let's look at the problem we are trying to solve: We
are trying to build a SOAP engine. In that world, a full Infoset
API is YAGNI. So, while the OM can easily be extended to cover
the full Infoset, one has to ask the question, "Why?". Making it
cover the full Infoset is not a requirement to solve the problem
we have to solve.
Later on, if someone were to offer a patch to "complete" the OM
then I'd have no problem. I do have a problem with us trying to
make the first snapshot of Axis2 have *anything* that is not
an absolutely required component of the solution.
Sanjiva.
Re: [Axis2] Re: Full infoset support
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
Sanjiva Weerawarana wrote:
>That's not true Dims .. SOAP doesn't *allow* you to transmit the
>full XML Infoset. So if you're doing SAAJ then you will *not*
>have the full Infoset present, even though the DOM behind SAAJ
>is capable of representing the full Infoset.
>
>The OM as it stands *is* capable of representing *any* XML that
>can appear inside <Body>...</Body>.
>
>
you mean API or implementation. I would like to make this distinction:
API projects XML Infoset view and implementation decides what is valid
in this view (for example: SOAP specific impl of XML Infoset PIs are
not allowed).
alek
>Sanjiva.
>
>----- Original Message -----
>From: "Davanum Srinivas" <da...@gmail.com>
>To: <ax...@ws.apache.org>; "Ajith Ranabahu" <aj...@gmail.com>
>Sent: Friday, December 03, 2004 6:49 PM
>Subject: Re: [Axis2] Re: Full infoset support
>
>
>
>
>>Ajith,
>>
>>Please remember, if the user wants to do doc/lit by hand....It implies
>>he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
>>So If you can't do complete XML Infoset there....then OM is useless
>>for him. Also remember you can't do SAAJ if OM does not support
>>complete XML Infoset.
>>
>>-- dims
>>
>>
>>On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
>><aj...@gmail.com> wrote:
>>
>>
>>>Hi there Alek,
>>>Sorry about the first mail which actually lacked the "Axis2" part. It
>>>was sent in by mistake.
>>>BTW i think the rationale behind making OM specific should be
>>>mentioned in a bit more detail here. This is how I see it so if other
>>>guys feel I am wrong please correct me.
>>>Ok here goes
>>>1. XML infoset is truly fundamental. No doubt about it. However that
>>>fundamentality only plays a role if the tool we are making is pure XML
>>>based. In this case our tool (Axis 2 to be specific) is focused into a
>>>specific subset of XML which is SOAP. So my point is that what we
>>>should be making is a SOAP object model (that is what OM is right
>>>now). You can just call it a bear-bones SOAP object model.
>>>2.We want this OM to be light weight. That is why we refrained from
>>>the easy choice, making OM on top of something like XOM or JDOM. We
>>>can incorporate the differed building capability to such object models
>>>without much hassel. However that wont solve our memory footprint
>>>problem. Hence OM was written purely from scratch to include the
>>>eseential information items and minimize the content in the memory.
>>>3. Well we DO gain a lot, in terms of memory,simplicity and speed by
>>>catering for SOAP only. The concern of many people is not another
>>>complete XML infoset representation (they already have plenty of
>>>choices like DOM,JDOM,XOM or even XML beans which has complete schema
>>>support). However there are very few SOAP stacks!
>>>
>>>Hope this helps to get our rationale through :)
>>>
>>>On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
>>>
>>>
>>><as...@cs.indiana.edu> wrote:
>>>
>>>
>>>>Eran Chinthaka wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>>>for me it is: AXIOM API must support XML Infoset and allow an API
>>>>>>>implementation that implements all XML infoset if it finds it
>>>>>>>
>>>>>>>
>necessary
>
>
>>>>>>>but default OM implementation is optimized for SOAP.
>>>>>>>
>>>>>>>i think Web Services are more than just SOAP and precluding use of
>>>>>>>XML/HTTP would be an error (especially that it is so easy to
>>>>>>>
>>>>>>>
>support).
>
>
>>>>>>>the only parts of XML Infoset that we may not want to support are
>>>>>>>
>>>>>>>
>those
>
>
>>>>>>>not required by SOAP (mostly related to PIs and DTDs as they are
>>>>>>>prohibited in SOAP 1.2 message for details see
>>>>>>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
>>>>>>>
>>>>>>>
>excluded? and
>
>
>>>>>>>comments are allowed just not inside message but in document ...)
>>>>>>>
>>>>>>>so think it can be simply done by having DTD related properties
>>>>>>>empty/default/null (such as those in
>>>>>>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl
>>>>>>>
>>>>>>>
>to
>
>
>>>>>>>throw exception if it sees DTD (configurable to do so as required
>>>>>>>
>>>>>>>
>for
>
>
>>>>>>>SOAP 1.2 ...)
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>[Chinthaka] can u all remember the basic concepts we agreed on the
>>>>>
>>>>>
>summit.
>
>
>>>>>One is YAGNI and one other is KISS. So if we just create some support
>>>>>
>>>>>
>for
>
>
>>>>>DTD and PI and just throw an exception, isn't it YAGNI or violating
>>>>>
>>>>>
>KISS
>
>
>>>>>???????
>>>>>
>>>>>
>>>>>
>>>>>
>>>>i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
>>>>something that can be simplified, omitted, or further KISSed.
>>>>
>>>>this is different from actual OM impls that may chose to not support
>>>>something in XML Infoset and throw exception (like in case of SOAP)
>>>>
>>>>therefore i do not think we differ much - i think XML Infoset API such
>>>>as AXIOM should be reflecting XML Infoset element information items
>>>>
>>>>
>and
>
>
>>>>properties but implementations can be optimized for subset of XML
>>>>Infoset (such as SOAP 1.2 or 1.1).
>>>>
>>>>however i do not see any considerable gains by fusing SOAP and XML
>>>>Infoset APIs together but it will make things bigger and more
>>>>complicated ...
>>>>
>>>>
>>>>
>>>>alek
>>>>
>>>>--
>>>>The best way to predict the future is to invent it - Alan Kay
>>>>
>>>>
>>>>
>>>>
>>>--
>>>Ajith Ranabahu
>>>
>>>
>>>
>>--
>>Davanum Srinivas - http://webservices.apache.org/~dims/
>>
>>
>
>
>
--
The best way to predict the future is to invent it - Alan Kay
Re: [Axis2] Re: Full infoset support
Posted by Sanjiva Weerawarana <sa...@opensource.lk>.
That's not true Dims .. SOAP doesn't *allow* you to transmit the
full XML Infoset. So if you're doing SAAJ then you will *not*
have the full Infoset present, even though the DOM behind SAAJ
is capable of representing the full Infoset.
The OM as it stands *is* capable of representing *any* XML that
can appear inside <Body>...</Body>.
Sanjiva.
----- Original Message -----
From: "Davanum Srinivas" <da...@gmail.com>
To: <ax...@ws.apache.org>; "Ajith Ranabahu" <aj...@gmail.com>
Sent: Friday, December 03, 2004 6:49 PM
Subject: Re: [Axis2] Re: Full infoset support
> Ajith,
>
> Please remember, if the user wants to do doc/lit by hand....It implies
> he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
> So If you can't do complete XML Infoset there....then OM is useless
> for him. Also remember you can't do SAAJ if OM does not support
> complete XML Infoset.
>
> -- dims
>
>
> On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
> <aj...@gmail.com> wrote:
> > Hi there Alek,
> > Sorry about the first mail which actually lacked the "Axis2" part. It
> > was sent in by mistake.
> > BTW i think the rationale behind making OM specific should be
> > mentioned in a bit more detail here. This is how I see it so if other
> > guys feel I am wrong please correct me.
> > Ok here goes
> > 1. XML infoset is truly fundamental. No doubt about it. However that
> > fundamentality only plays a role if the tool we are making is pure XML
> > based. In this case our tool (Axis 2 to be specific) is focused into a
> > specific subset of XML which is SOAP. So my point is that what we
> > should be making is a SOAP object model (that is what OM is right
> > now). You can just call it a bear-bones SOAP object model.
> > 2.We want this OM to be light weight. That is why we refrained from
> > the easy choice, making OM on top of something like XOM or JDOM. We
> > can incorporate the differed building capability to such object models
> > without much hassel. However that wont solve our memory footprint
> > problem. Hence OM was written purely from scratch to include the
> > eseential information items and minimize the content in the memory.
> > 3. Well we DO gain a lot, in terms of memory,simplicity and speed by
> > catering for SOAP only. The concern of many people is not another
> > complete XML infoset representation (they already have plenty of
> > choices like DOM,JDOM,XOM or even XML beans which has complete schema
> > support). However there are very few SOAP stacks!
> >
> > Hope this helps to get our rationale through :)
> >
> > On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
> >
> >
> > <as...@cs.indiana.edu> wrote:
> > > Eran Chinthaka wrote:
> > >
> > >
> > >
> > > >>>for me it is: AXIOM API must support XML Infoset and allow an API
> > > >>>implementation that implements all XML infoset if it finds it
necessary
> > > >>>but default OM implementation is optimized for SOAP.
> > > >>>
> > > >>>i think Web Services are more than just SOAP and precluding use of
> > > >>>XML/HTTP would be an error (especially that it is so easy to
support).
> > > >>>
> > > >>>the only parts of XML Infoset that we may not want to support are
those
> > > >>>not required by SOAP (mostly related to PIs and DTDs as they are
> > > >>>prohibited in SOAP 1.2 message for details see
> > > >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more
excluded? and
> > > >>>comments are allowed just not inside message but in document ...)
> > > >>>
> > > >>>so think it can be simply done by having DTD related properties
> > > >>>empty/default/null (such as those in
> > > >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl
to
> > > >>>throw exception if it sees DTD (configurable to do so as required
for
> > > >>>SOAP 1.2 ...)
> > > >>>
> > > >>>
> > > >
> > > >[Chinthaka] can u all remember the basic concepts we agreed on the
summit.
> > > >One is YAGNI and one other is KISS. So if we just create some support
for
> > > >DTD and PI and just throw an exception, isn't it YAGNI or violating
KISS
> > > >???????
> > > >
> > > >
> > > i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
> > > something that can be simplified, omitted, or further KISSed.
> > >
> > > this is different from actual OM impls that may chose to not support
> > > something in XML Infoset and throw exception (like in case of SOAP)
> > >
> > > therefore i do not think we differ much - i think XML Infoset API such
> > > as AXIOM should be reflecting XML Infoset element information items
and
> > > properties but implementations can be optimized for subset of XML
> > > Infoset (such as SOAP 1.2 or 1.1).
> > >
> > > however i do not see any considerable gains by fusing SOAP and XML
> > > Infoset APIs together but it will make things bigger and more
> > > complicated ...
> > >
> > >
> > >
> > > alek
> > >
> > > --
> > > The best way to predict the future is to invent it - Alan Kay
> > >
> > >
> >
> >
> > --
> > Ajith Ranabahu
> >
>
>
> --
> Davanum Srinivas - http://webservices.apache.org/~dims/
Re: [Axis2] Re: Full infoset support
Posted by Davanum Srinivas <da...@gmail.com>.
Ajith,
Please remember, if the user wants to do doc/lit by hand....It implies
he wants to do EVERYTHING possible with XML inside the SOAP Envelope.
So If you can't do complete XML Infoset there....then OM is useless
for him. Also remember you can't do SAAJ if OM does not support
complete XML Infoset.
-- dims
On Fri, 3 Dec 2004 16:20:20 +0600, Ajith Ranabahu
<aj...@gmail.com> wrote:
> Hi there Alek,
> Sorry about the first mail which actually lacked the "Axis2" part. It
> was sent in by mistake.
> BTW i think the rationale behind making OM specific should be
> mentioned in a bit more detail here. This is how I see it so if other
> guys feel I am wrong please correct me.
> Ok here goes
> 1. XML infoset is truly fundamental. No doubt about it. However that
> fundamentality only plays a role if the tool we are making is pure XML
> based. In this case our tool (Axis 2 to be specific) is focused into a
> specific subset of XML which is SOAP. So my point is that what we
> should be making is a SOAP object model (that is what OM is right
> now). You can just call it a bear-bones SOAP object model.
> 2.We want this OM to be light weight. That is why we refrained from
> the easy choice, making OM on top of something like XOM or JDOM. We
> can incorporate the differed building capability to such object models
> without much hassel. However that wont solve our memory footprint
> problem. Hence OM was written purely from scratch to include the
> eseential information items and minimize the content in the memory.
> 3. Well we DO gain a lot, in terms of memory,simplicity and speed by
> catering for SOAP only. The concern of many people is not another
> complete XML infoset representation (they already have plenty of
> choices like DOM,JDOM,XOM or even XML beans which has complete schema
> support). However there are very few SOAP stacks!
>
> Hope this helps to get our rationale through :)
>
> On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
>
>
> <as...@cs.indiana.edu> wrote:
> > Eran Chinthaka wrote:
> >
> >
> >
> > >>>for me it is: AXIOM API must support XML Infoset and allow an API
> > >>>implementation that implements all XML infoset if it finds it necessary
> > >>>but default OM implementation is optimized for SOAP.
> > >>>
> > >>>i think Web Services are more than just SOAP and precluding use of
> > >>>XML/HTTP would be an error (especially that it is so easy to support).
> > >>>
> > >>>the only parts of XML Infoset that we may not want to support are those
> > >>>not required by SOAP (mostly related to PIs and DTDs as they are
> > >>>prohibited in SOAP 1.2 message for details see
> > >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more excluded? and
> > >>>comments are allowed just not inside message but in document ...)
> > >>>
> > >>>so think it can be simply done by having DTD related properties
> > >>>empty/default/null (such as those in
> > >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl to
> > >>>throw exception if it sees DTD (configurable to do so as required for
> > >>>SOAP 1.2 ...)
> > >>>
> > >>>
> > >
> > >[Chinthaka] can u all remember the basic concepts we agreed on the summit.
> > >One is YAGNI and one other is KISS. So if we just create some support for
> > >DTD and PI and just throw an exception, isn't it YAGNI or violating KISS
> > >???????
> > >
> > >
> > i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
> > something that can be simplified, omitted, or further KISSed.
> >
> > this is different from actual OM impls that may chose to not support
> > something in XML Infoset and throw exception (like in case of SOAP)
> >
> > therefore i do not think we differ much - i think XML Infoset API such
> > as AXIOM should be reflecting XML Infoset element information items and
> > properties but implementations can be optimized for subset of XML
> > Infoset (such as SOAP 1.2 or 1.1).
> >
> > however i do not see any considerable gains by fusing SOAP and XML
> > Infoset APIs together but it will make things bigger and more
> > complicated ...
> >
> >
> >
> > alek
> >
> > --
> > The best way to predict the future is to invent it - Alan Kay
> >
> >
>
>
> --
> Ajith Ranabahu
>
--
Davanum Srinivas - http://webservices.apache.org/~dims/
Re: [Axis2] Re: Full infoset support
Posted by Ajith Ranabahu <aj...@gmail.com>.
Hi there Alek,
Sorry about the first mail which actually lacked the "Axis2" part. It
was sent in by mistake.
BTW i think the rationale behind making OM specific should be
mentioned in a bit more detail here. This is how I see it so if other
guys feel I am wrong please correct me.
Ok here goes
1. XML infoset is truly fundamental. No doubt about it. However that
fundamentality only plays a role if the tool we are making is pure XML
based. In this case our tool (Axis 2 to be specific) is focused into a
specific subset of XML which is SOAP. So my point is that what we
should be making is a SOAP object model (that is what OM is right
now). You can just call it a bear-bones SOAP object model.
2.We want this OM to be light weight. That is why we refrained from
the easy choice, making OM on top of something like XOM or JDOM. We
can incorporate the differed building capability to such object models
without much hassel. However that wont solve our memory footprint
problem. Hence OM was written purely from scratch to include the
eseential information items and minimize the content in the memory.
3. Well we DO gain a lot, in terms of memory,simplicity and speed by
catering for SOAP only. The concern of many people is not another
complete XML infoset representation (they already have plenty of
choices like DOM,JDOM,XOM or even XML beans which has complete schema
support). However there are very few SOAP stacks!
Hope this helps to get our rationale through :)
On Fri, 03 Dec 2004 03:08:57 -0500, Aleksander Slominski
<as...@cs.indiana.edu> wrote:
> Eran Chinthaka wrote:
>
>
>
> >>>for me it is: AXIOM API must support XML Infoset and allow an API
> >>>implementation that implements all XML infoset if it finds it necessary
> >>>but default OM implementation is optimized for SOAP.
> >>>
> >>>i think Web Services are more than just SOAP and precluding use of
> >>>XML/HTTP would be an error (especially that it is so easy to support).
> >>>
> >>>the only parts of XML Infoset that we may not want to support are those
> >>>not required by SOAP (mostly related to PIs and DTDs as they are
> >>>prohibited in SOAP 1.2 message for details see
> >>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more excluded? and
> >>>comments are allowed just not inside message but in document ...)
> >>>
> >>>so think it can be simply done by having DTD related properties
> >>>empty/default/null (such as those in
> >>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl to
> >>>throw exception if it sees DTD (configurable to do so as required for
> >>>SOAP 1.2 ...)
> >>>
> >>>
> >
> >[Chinthaka] can u all remember the basic concepts we agreed on the summit.
> >One is YAGNI and one other is KISS. So if we just create some support for
> >DTD and PI and just throw an exception, isn't it YAGNI or violating KISS
> >???????
> >
> >
> i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
> something that can be simplified, omitted, or further KISSed.
>
> this is different from actual OM impls that may chose to not support
> something in XML Infoset and throw exception (like in case of SOAP)
>
> therefore i do not think we differ much - i think XML Infoset API such
> as AXIOM should be reflecting XML Infoset element information items and
> properties but implementations can be optimized for subset of XML
> Infoset (such as SOAP 1.2 or 1.1).
>
> however i do not see any considerable gains by fusing SOAP and XML
> Infoset APIs together but it will make things bigger and more
> complicated ...
>
>
>
> alek
>
> --
> The best way to predict the future is to invent it - Alan Kay
>
>
--
Ajith Ranabahu
Re: [Axis2] Re: Full infoset support
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
Eran Chinthaka wrote:
>>>for me it is: AXIOM API must support XML Infoset and allow an API
>>>implementation that implements all XML infoset if it finds it necessary
>>>but default OM implementation is optimized for SOAP.
>>>
>>>i think Web Services are more than just SOAP and precluding use of
>>>XML/HTTP would be an error (especially that it is so easy to support).
>>>
>>>the only parts of XML Infoset that we may not want to support are those
>>>not required by SOAP (mostly related to PIs and DTDs as they are
>>>prohibited in SOAP 1.2 message for details see
>>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more excluded? and
>>>comments are allowed just not inside message but in document ...)
>>>
>>>so think it can be simply done by having DTD related properties
>>>empty/default/null (such as those in
>>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl to
>>>throw exception if it sees DTD (configurable to do so as required for
>>>SOAP 1.2 ...)
>>>
>>>
>
>[Chinthaka] can u all remember the basic concepts we agreed on the summit.
>One is YAGNI and one other is KISS. So if we just create some support for
>DTD and PI and just throw an exception, isn't it YAGNI or violating KISS
>???????
>
>
i regard XML infoset API as _fundamental_ (so it is YAGNI) and not
something that can be simplified, omitted, or further KISSed.
this is different from actual OM impls that may chose to not support
something in XML Infoset and throw exception (like in case of SOAP)
therefore i do not think we differ much - i think XML Infoset API such
as AXIOM should be reflecting XML Infoset element information items and
properties but implementations can be optimized for subset of XML
Infoset (such as SOAP 1.2 or 1.1).
however i do not see any considerable gains by fusing SOAP and XML
Infoset APIs together but it will make things bigger and more
complicated ...
alek
--
The best way to predict the future is to invent it - Alan Kay
RE: [Axis2] Re: Full infoset support
Posted by Eran Chinthaka <ch...@opensource.lk>.
>>
>>i know i may be PITA but please do put [Axis2] somewhere in email
>>subject ...
[Chinthaka] Alek, Axis2 was there in the subject. My email filters got it
correct.
>>
>>my reply is below.
>>
>>Ajith Ranabahu wrote:
>>
>>>Since this issue was taken up in the last IRC session and ended with
>>>no conclusion, I suppose we should talks about this seriously and take
>>>a decision. So I am going to put forward the qeustion in simple terms
>>>and put up my answer with reasons.
>>>
>>>Q - Do we need full XML infoset support with OM?
>>>
>>>A - No.
>>>
>>>
>>please define what is exactly full XML infoset support?
>>
>>for me it is: AXIOM API must support XML Infoset and allow an API
>>implementation that implements all XML infoset if it finds it necessary
>>but default OM implementation is optimized for SOAP.
>>
>>i think Web Services are more than just SOAP and precluding use of
>>XML/HTTP would be an error (especially that it is so easy to support).
>>
>>the only parts of XML Infoset that we may not want to support are those
>>not required by SOAP (mostly related to PIs and DTDs as they are
>>prohibited in SOAP 1.2 message for details see
>>http://www.w3.org/TR/soap12-part1/#soapenv - anything more excluded? and
>>comments are allowed just not inside message but in document ...)
>>
>>so think it can be simply done by having DTD related properties
>>empty/default/null (such as those in
>>http://www.w3.org/TR/xml-infoset/#infoitem.document) and AXIOM impl to
>>throw exception if it sees DTD (configurable to do so as required for
>>SOAP 1.2 ...)
[Chinthaka] can u all remember the basic concepts we agreed on the summit.
One is YAGNI and one other is KISS. So if we just create some support for
DTD and PI and just throw an exception, isn't it YAGNI or violating KISS
???????
Thankx,
Eran Chinthaka
>>
>>this is very quick analysis - let me know if i missed something.
>>
>>thanks,
>>
>>alek
>>
>>--
>>The best way to predict the future is to invent it - Alan Kay