[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [wsbpel] Issue 11 - Proposal For Vote
I think that COPY has some minimal utility for a few key scenarios but I
do not believe it is or was ever intended to be a generic data
manipulation language. Copy, as Satish has pointed out, was really
intended to just enable a few very limited scenarios having to do with
things like assigning partnerLinks EPRs and copying message parts.
As for XUPDATE, either they will take forever because updating XML is
hard and we should therefore not be taking it on ourselves. Or, they
won't take forever because the subject is tractable, in which case we
should let them get on with it and not take it on ourselves.
Also, XUPDATE isn't the only language available that can manipulate
arbitrary XML structures. XSLT, Java, Python, C, C# and so on are all
able to manipulate arbitrary XML structures. It is preferable to define
bindings to those languages than to turn BPEL into a generic programming
language. As for where to standardize such bindings? There are many
choices but OASIS would be an obvious one.
Thanks,
Yaron
Ugo Corda wrote:
> > Its intention was strictly to address 'programming
> > in the large' issues, which does not include data
> > manipulation.
>
> There is already a lot of data manipulation in BPEL. Are you suggesting
> that it should be removed? Or do you think that what is already there
> satisfies the 80/20 rule? I certainly don't.
>
> > It would be, in my opinion, much better were we to take an existing
> > language that supports XML manipulation and provide a
> > standard binding for it to BPEL.
>
> Who is going to provide this standard binding? You say it is out of
> scope for the TC. So who else would do it? Are you thinking of XUpdate?
> That activity basically just started from scratch once again last
> February. It might take another couple of years to finalize (judging
> from the time taken by XQuery so far). Standardizing an XML manipulation
> language for BPEL a couple of years from now will be completely futile:
> all possible damage to BPEL portability will already have been done by
> then.
>
> Ugo
>
> > -----Original Message-----
> > From: Yaron Y. Goland [mailto:ygoland@bea.com]
> > Sent: Monday, April 04, 2005 6:34 PM
> > To: Ugo Corda
> > Cc: Satish Thatte; wsbpel@lists.oasis-open.org
> > Subject: Re: [wsbpel] Issue 11 - Proposal For Vote
> >
> >
> > BPEL is not nor was it ever intended to be a universal generic
> > programming language. Its intention was strictly to address
> > 'programming
> > in the large' issues, which does not include data
> > manipulation. I think
> > it inappropriate to bloat BPEL with features that are not relevant to
> > its core mission.
> >
> > It would be, in my opinion, much better were we to take an existing
> > language that supports XML manipulation and provide a
> > standard binding
> > for it to BPEL. Although such work is clearly outside the
> > scope of the
> > TC it is a scenario that BPEL explicitly provides extension hooks to
> > support.
> >
> > Yaron
> >
> > Ugo Corda wrote:
> > > While you worry about the slippery slope, I am even more concerned
> > > about the portability problems caused by lack of support
> > for the most
> > > elementary assignment capabilities, like being able to
> > incrementally
> > > add instances of a repeating element inside a loop.
> > >
> > > This threat to portability is quite real. As far as I know, most
> > > existing BPEL implementations have already introduced their own
> > > proprietary extensions in order to overcome these limitations. For
> > > what I can see, this area stands out as the most likely source of
> > > serious portability problems for BPEL.
> > >
> > > Ugo
> > >
> > > > -----Original Message-----
> > > > From: Satish Thatte [mailto:satisht@microsoft.com]
> > > > Sent: Sunday, April 03, 2005 11:36 PM
> > > > To: Ugo Corda
> > > > Cc: wsbpel@lists.oasis-open.org
> > > > Subject: RE: [wsbpel] Issue 11 - Proposal For Vote
> > > >
> > > >
> > > > The copy operation is minimally needed to construct WSDL
> > > > messages out of parts, and to assign partnerLinks and
> > > > properties, which are core BPEL concepts. I agree with
> > you > that
> > > strict consistentcy requires we remove other options but > we have
> > > chosen to take one simple step beyond but with the >
> > principle that
> > > we do not allow changing the XML structure of > the
> > destination of an
> > > assignment. Thus query-based > assignment is really the same as
> > > property assignment but > without explicit property definition. I
> > > find this a good > place to stop to avoid stepping on a
> > really steep
> > > slippery slope. >
> > > > ________________________________
> > > >
> > > > From: Ugo Corda [mailto:UCorda@SeeBeyond.com]
> > > > Sent: Sun 4/3/2005 10:28 PM
> > > > To: Satish Thatte
> > > > Cc: wsbpel@lists.oasis-open.org
> > > > Subject: RE: [wsbpel] Issue 11 - Proposal For Vote
> > > >
> > > >
> > > >
> > > > I would better understand your position if you were also
> > > > proposing to remove the "copy" operation from the language.
> > > > Why leave a "copy" operation with grossly crippled functionality?
> > > >
> > > > Ugo
> > > >
> > > > > -----Original Message-----
> > > > > From: Satish Thatte [mailto:satisht@microsoft.com]
> > > > > Sent: Sunday, April 03, 2005 10:06 PM
> > > > > To: Ugo Corda; ygoland@bea.com
> > > > > Cc: wsbpel@lists.oasis-open.org
> > > > > Subject: RE: [wsbpel] Issue 11 - Proposal For Vote
> > > > >
> > > > >
> > > > > Regardless of language, I still don't understand why it is
> > > > useful to
> > > > > add a small subset of the DOM API in BPEL statement
> > form. While I
> > > > > understand the desire for completeness, we need to
> > > > recognize that BPEL
> > > > > is in fact not a complete language. This seems to me
> > like a (very)
> > > > > partial solution to the problem of XML data manipulation in
> > > > processes,
> > > > > which in fact is outside our scope and hence its solution
> > > > is best left
> > > > > outside the language. Web service invocations already act as
> > > > > completely general XML transducers in principle and we will
> > > > > have enough extensibility to invent new transducers using
> > > > > other technologies.
> > > > >
> > > > > ________________________________
> > > > >
> > > > > From: Ugo Corda [mailto:UCorda@SeeBeyond.com]
> > > > > Sent: Thu 3/31/2005 2:37 PM
> > > > > To: ygoland@bea.com
> > > > > Cc: wsbpel@lists.oasis-open.org
> > > > > Subject: RE: [wsbpel] Issue 11 - Proposal For Vote
> > > > >
> > > > >
> > > > >
> > > > > Well, let's look, for example, at the latest version of
> > > > this section
> > > > > 14.3 paragraph (from the resolution to issue 103):
> > > > >
> > > > > "The from-spec and to-spec MUST yield a node-set that
> > > > contains exactly
> > > > > one node. If the from-spec or to-spec selects zero nodes or
> > > > more than
> > > > > one node during execution, then the standard fault
> > > > > bpws:selectionFailure MUST be thrown by a compliant
> > implementation".
> > > > >
> > > > > I don't see any generic infoset terminology in that
> > > > paragraph either.
> > > > >
> > > > > Ugo
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Yaron Y. Goland [mailto:ygoland@bea.com]
> > > > > > Sent: Thursday, March 31, 2005 1:53 PM
> > > > > > To: Ugo Corda
> > > > > > Cc: wsbpel@lists.oasis-open.org
> > > > > > Subject: Re: [wsbpel] Issue 11 - Proposal For Vote
> > > > > >
> > > > > >
> > > > > > This proposal does not use infoset terminology. In fact,
> > > > > the proposal
> > > > > > looks to be XPATH specific. Defining issue 11 in
> > terms of XPATH
> > > > > > semantics would mean that the issue 11 features
> > could change in
> > > > > > functionality based on which expression languages were
> > > > used as the
> > > > > > source or destination. Minimally a proposal to
> > resolve this issue
> > > > > > should be written using generic infoset terminology so
> > > > that issue 11
> > > > > > behavior will be consistent regardless of expression
> > language.
> > > > > >
> > > > > > Thanks,
> > > > > >
> > > > > > Yaron
> > > > > >
> > > > > > Ugo Corda wrote:
> > > > > > > I have been working on a resolution of this issue
> > with a few TC
> > > > > > > members who are interested in a type of resolution like the
> > > > > > one that
> > > > > > > Danny presented a few months ago. Please find our current
> > > > > proposal
> > > > > > > attached.
> > > > > > >
> > > > > > > Regards,
> > > > > > > Ugo
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > ----------------------------------------------------------------------
> > > > > > > --
> > > > > > >
> > > > > > >
> > > > > > > Issue 11 - Proposal
> > > > > > >
> > > > > > >
> > > > > > > 1. Append
> > > > > > >
> > > > > > > Syntax:
> > > > > > > <bpel:assign>
> > > > > > > <bpel:append>
> > > > > > > <bpel:from ... />
> > > > > > > <bpel:to ... />
> > > > > > > </bpel:append>
> > > > > > > </bpel:assign>
> > > > > > >
> > > > > > > The from-spec within <append> yields a single node or a
> > > > > > node set. Note
> > > > > > > that the
> > > > > > > from-spec of <bpel:copy> still MUST yield ONLY one node.
> > > > > > The node set will be
> > > > > > > processed in document order (unless an alternative order is
> > > > > > specified in the
> > > > > > > underlying query language). If the from-spec yields atomic
> > > > > > values (e.g.
> > > > > > > numbers), those atomic values can be converted to text
> > > > > > nodes, when needed. If
> > > > > > > the from-spec yields zero node, "bpel:selectionFailure"
> > > > > > fault MUST be thrown. If
> > > > > > > the from-spec yields an attribute node, then
> > > > > > "bpel:selectionFailure" fault MUST
> > > > > > > be thrown.
> > > > > > >
> > > > > > > {{ Background context note on XPath 1.0 and XSLT 1.0: the
> > > > > > selected set
> > > > > > > of nodes
> > > > > > > by XPath 1.0 in XSLT 1.0 is processed in document order.
> > > > > > Hence, it is feasible
> > > > > > > and natural to process the nodes selected by the from-spec
> > > > > > in the document order
> > > > > > > in <append> and other similar XML data operations. For
> > > > > > details, please see
> > > > > > > section 5.4 in [XSLT1.0] specification. }}
> > > > > > >
> > > > > > > The to-spec MUST yield one single element node as the
> > > > > > target element
> > > > > > > node.
> > > > > > > Otherwise, "bpel:selectionFailure" fault will be generated.
> > > > > > The to-spec cannot
> > > > > > > refer to a partnerLink. The node-set returned by the
> > > > > > from-spec will be appended
> > > > > > > orderly as child nodes to the target element node.
> > > > > > >
> > > > > > >
> > > > > > > 2. Insert
> > > > > > >
> > > > > > >
> > > > > > > 2.1 insertBefore
> > > > > > >
> > > > > > > Syntax:
> > > > > > > <bpel:assign>
> > > > > > > <bpel:insertBefore>
> > > > > > > <bpel:from ... />
> > > > > > > <bpel:to ... />
> > > > > > > </bpel:insertBefore>
> > > > > > > </bpel:assign>
> > > > > > >
> > > > > > > The restriction and semantics of from-spec under
> > > > insertBefore is
> > > > > > > similar to those in the case of <bpel:append>.
> > > > > > >
> > > > > > > The to-spec under <insertBefore> MUST points to one or more
> > > > > > nodes. If
> > > > > > > more than
> > > > > > > one nodes are returned, the first node will be used as the
> > > > > > reference node. The
> > > > > > > word "first" here means respect to the order of the node
> > > > > > set selected by
> > > > > > > to-spec, which is by default in document order, unless an
> > > > > > alternative order is
> > > > > > > specified in the underlying query language.
> > > > > > >
> > > > > > > The reference node MUST be an element node. The
> > parent of the
> > > > > > > reference node MUST be an element node also. Otherwise,
> > > > > > "bpel:selectionFailure" fault will be
> > > > > > > generated. The to-spec cannot refer to a partnerLink.
> > > > > > >
> > > > > > > The node set generated by the from-spec will be inserted
> > > > > before the
> > > > > > > reference node in the document order (unless an
> > > > > alternative order is
> > > > > > specified in the
> > > > > > > underlying query language).
> > > > > > >
> > > > > > >
> > > > > > > 2.2 insertAfter
> > > > > > >
> > > > > > >
> > > > > > > Syntax:
> > > > > > > <bpel:assign>
> > > > > > > <bpel:insertAfter>
> > > > > > > <bpel:from ... />
> > > > > > > <bpel:to ... />
> > > > > > > </bpel:insertAfter>
> > > > > > > </bpel:assign>
> > > > > > >
> > > > > > > <insertAfter> is very similar to <insertBefore>. Except:
> > > > > > >
> > > > > > > * If more than one nodes are returned by the to-spec,
> > > > > > the last node will be
> > > > > > > used as the reference node. The word "last" here
> > > > > > means respect to the
> > > > > > > order of the node set selected by to-spec, which is
> > > > > > by default in document
> > > > > > > order, unless an alternative order is specified in
> > > > > > the underlying query
> > > > > > > language.
> > > > > > > * Instead of inserting nodes before the reference node,
> > > > > > the nodes selected
> > > > > > > by will be inserted after the reference node in the
> > > > > > document order (unless
> > > > > > > an alternative order is specified in the underlying
> > > > > > query language).
> > > > > > > * This operation can also be considered a macro of
> > > > > > conditional-switch +
> > > > > > > (append or insertBefore).
> > > > > > >
> > > > > > >
> > > > > > > 3. Remove
> > > > > > >
> > > > > > > Syntax:
> > > > > > > <bpel:assign>
> > > > > > > <bpel:remove>
> > > > > > > <bpel:target ... />
> > > > > > > </bpel:append>
> > > > > > > </bpel:assign>
> > > > > > >
> > > > > > > The syntax of "bpel:target" is similar to and a subset of
> > > > > > to-spec used
> > > > > > > in <copy>
> > > > > > > with the "partnerLink" attribute removed. Similarly, XPath
> > > > > > 1.0 is used as the
> > > > > > > default query language for <target>-spec.
> > > > > > >
> > > > > > > This <remove> operation remove nodes specified by the
> > > > > <target>-spec
> > > > > > > from their parent nodes. Nodes specified by
> > <target>-spec MAY be
> > > > > > multiple. Nodes being
> > > > > > > removed can be: text nodes, attribute nodes and
> > element nodes.
> > > > > > >
> > > > > > > If the <target>-spec returns zero node, then
> > > > > > "bpel:selectionFailure"
> > > > > > > fault MUST
> > > > > > > be thrown.
> > > > > > >
> > > > > > >
> > > > > > > Reference:
> > > > > > >
> > > > > > > * [XSLT1.0] http://www.w3.org/TR/xslt
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > [date: 2005-03-30]
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > ----------------------------------------------------------------------
> > > > > > > --
> > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > ---------------------------------------------------------------------
> > > > > > > To unsubscribe, e-mail:
> > wsbpel-unsubscribe@lists.oasis-open.org
> > > > > > > For additional commands, e-mail:
> > > > wsbpel-help@lists.oasis-open.org
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > ---------------------------------------------------------------------
> > > > > To unsubscribe from this mail list, you must leave the
> > > > OASIS TC that
> > > > > generates this mail. You may a link to this group and all
> > > > your TCs in
> > > > > OASIS
> > > > > at:
> > > > >
> > > >
> > https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > > >
> > > >
> > >
> > >
> > ---------------------------------------------------------------------
> > > To unsubscribe from this mail list, you must leave the
> > OASIS TC that
> > > generates this mail. You may a link to this group and all
> > your TCs in
> > > OASIS
> > > at:
> > >
> > https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
> > >
> >
> >
>
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]