Home » Modeling » EMF » xsd:any I'm clearly not using it correctly...
|
| Re: xsd:any I'm clearly not using it correctly... [message #424588 is a reply to message #424587] |
Thu, 30 October 2008 18:27   |
 Ed Merks
Messages: 33142
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------040403090102050403060004
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Jason,
Comments below.
jason henriksen wrote:
>
> Using this XSD I generate an object.
>
> <xsd:element name="setOutputMessage">
>
> <xsd:complexType>
> <xsd:sequence>
> <xsd:element name="tkiid" type="xsd:string"/>
> <xsd:any maxOccurs="1" minOccurs="1" processContents="lax"/>
> </xsd:sequence>
> </xsd:complexType>
> </xsd:element>
>
> I try to put an emf object in and I get a class cast exception:
>
> ExecuteHTResponseType test = new ExecuteHTResponseTypeImpl();
> SetOutputMessageType somt = new SetOutputMessageTypeImpl();
That makes me sad. Why aren't you using factories?
> somt.getAny().add(0, test);
You're not using Java 5.0 that it complains?
>
> Doing this gives a class cast because it tries to put the 'test'
> object into the tkiid field.
I would imagine it's trying to add an FeatureMap.Entry to the list...
> So clearly I'm not getting how the Sequence
Sequence? Is this an SDO thing?
> coming back from somt.getAny() is supposed to be used.
>
> All I need to do is insert the object after the tkiid tag when the
> object is serialized. Can you offer any example code?
This article will give you a better idea of who the FeatureMaps work
Binding XML to Java
< http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava>
Personally I think SDO sucks and index-based features as a user API are
among the suckiest thing I could imagine. They're totally error prone.
Putting them on Sequence is an abomination, given that Sequences are
used to deal with wildcards where a feature ID can only be mapped to a
feature in the object which can't possibly be the desired thing. There
is no magical correct number you can pass to that API for this use
case. You'll need to find a Property that corresponds to a global element.
I think I'm going to get rid of SDO:
<https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402>
https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402
You'd be better not to use it, in my opinion.
>
> Thanks,
>
> Jason Henriksen
--------------040403090102050403060004
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Jason,<br>
<br>
Comments below.<br>
<br>
<br>
jason henriksen wrote:
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite"><br>
Using this XSD I generate an object.
<br>
<br>
<xsd:element name="setOutputMessage">
<br>
<br>
<xsd:complexType>
<br>
<xsd:sequence>
<br>
<xsd:element name="tkiid" type="xsd:string"/>
<br>
<xsd:any maxOccurs="1" minOccurs="1" processContents="lax"/>
<br>
</xsd:sequence>
<br>
</xsd:complexType>
<br>
</xsd:element>
<br>
<br>
I try to put an emf object in and I get a class cast exception:
<br>
<br>
ExecuteHTResponseTyp e test = new
ExecuteHTResponseTypeImpl(); &a mp;nbsp; <br>
SetOutputMessageType somt = new SetOutputMessageTypeImpl();
<br>
</blockquote>
That makes me sad. Why aren't you using factories?<br>
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite"> somt.getAny().add(0,
test);
<br>
</blockquote>
You're not using Java 5.0 that it complains?<br>
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite"><br>
Doing this gives a class cast because it tries to put the 'test' object
into the tkiid field.</blockquote>
I would imagine it's trying to add an FeatureMap.Entry to the list...<br>
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite"> So
clearly I'm not getting how the Sequence </blockquote>
Sequence? Is this an SDO thing?<br>
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite">coming
back from somt.getAny() is supposed to be used.
<br>
<br>
All I need to do is insert the object after the tkiid tag when the
object is serialized. Can you offer any example code?
<br>
</blockquote>
This article will give you a better idea of who the FeatureMaps work<br>
<blockquote><a target="_out"
href=" http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava">Binding
XML to Java</a><br>
</blockquote>
Personally I think SDO sucks and index-based features as a user API are
among the suckiest thing I could imagine. They're totally error
prone. Putting them on Sequence is an abomination, given that
Sequences are used to deal with wildcards where a feature ID can only
be mapped to a feature in the object which can't possibly be the
desired thing. There is no magical correct number you can pass to
that API for this use case. You'll need to find a Property that
corresponds to a global element.<br>
<br>
I think I'm going to get rid of SDO:<a
href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402"><br>
</a>
<blockquote><a
href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402">https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402</a><br>
</blockquote>
You'd be better not to use it, in my opinion.<br>
<blockquote cite="mid:gecsvn$58h$1@build.eclipse.org" type="cite"><br>
Thanks,
<br>
<br>
Jason Henriksen
<br>
</blockquote>
</body>
</html>
--------------040403090102050403060004--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
|
| Re: xsd:any I'm clearly not using it correctly... [message #424590 is a reply to message #424588] |
Thu, 30 October 2008 18:54 |
Jason Henriksen
Messages: 231
Registered: July 2009 |
Senior Member |
|
|
>> I try to put an emf object in and I get a class cast exception:
>>
>> ExecuteHTResponseType test = new ExecuteHTResponseTypeImpl();
>> SetOutputMessageType somt = new SetOutputMessageTypeImpl();
> That makes me sad. Why aren't you using factories?
They're just verbose. I love factories as a pattern for building code
that may need alteration at runtime. But I can't touch production
without a full recompile/QA/Acceptance flow anyway so why not just say
what I mean? The majority of my objects are business rule
implementations that only make sense in one place anyway. I just make
the objects have public constructors so that I can just get my work done.
If there's any hope that I factory will be useful, I'll leave it. But
if I'm writing sample code, one off rules or other off-the-cuff stuff, I
just cut to the chase and use the constructor.
>> somt.getAny().add(0, test);
> You're not using Java 5.0 that it complains?
We are using Java 5.
>> Doing this gives a class cast because it tries to put the 'test'
>> object into the tkiid field.
> I would imagine it's trying to add an FeatureMap.Entry to the list...
>> So clearly I'm not getting how the Sequence
> Sequence? Is this an SDO thing?
Yes, it is an SDO. The fact that EMF works as SDOs is one of the
selling points of us using it since it will be compatible with IBMs
other SDO tooling.
>> coming back from somt.getAny() is supposed to be used.
>>
>> All I need to do is insert the object after the tkiid tag when the
>> object is serialized. Can you offer any example code?
> This article will give you a better idea of who the FeatureMaps work
>
> Binding XML to Java
> < http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava>
Thanks, I'll look at that and give it a try.
>
> Personally I think SDO sucks and index-based features as a user API are
> among the suckiest thing I could imagine. They're totally error prone.
> Putting them on Sequence is an abomination, given that Sequences are
> used to deal with wildcards where a feature ID can only be mapped to a
> feature in the object which can't possibly be the desired thing. There
> is no magical correct number you can pass to that API for this use
> case. You'll need to find a Property that corresponds to a global element.
>
> I think I'm going to get rid of SDO:
> <https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402>
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402
>
> You'd be better not to use it, in my opinion.
It's probably all 'marketecture' but the SDO badge on EMF makes my
client accept it. (They're a VERY pro-IBM shop). I'm worried that if
you remove the SDO stuff they'll either push to replace EMF with the SDO
stuff from Tuscany or push back on the use of EMF all together. :(
BTW, I agree with you completely about the index numbers being a really
bad idea. I just know that I had to refer to EMF object instances as
SDOs for weeks before they warmed up to the idea of EMF.
>>
>> Thanks,
>>
>> Jason Henriksen
|
|
|
| Re: xsd:any I'm clearly not using it correctly... [message #424591 is a reply to message #424590] |
Thu, 30 October 2008 19:02 |
Ed Merks
Messages: 33142
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------030903090702000508000709
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Jason,
Comments below.
jason henriksen wrote:
>>> I try to put an emf object in and I get a class cast exception:
>>>
>>> ExecuteHTResponseType test = new
>>> ExecuteHTResponseTypeImpl(); SetOutputMessageType somt =
>>> new SetOutputMessageTypeImpl();
>> That makes me sad. Why aren't you using factories?
>
> They're just verbose. I love factories as a pattern for building code
> that may need alteration at runtime. But I can't touch production
> without a full recompile/QA/Acceptance flow anyway so why not just say
> what I mean? The majority of my objects are business rule
> implementations that only make sense in one place anyway. I just make
> the objects have public constructors so that I can just get my work done.
>
> If there's any hope that I factory will be useful, I'll leave it. But
> if I'm writing sample code, one off rules or other off-the-cuff stuff,
> I just cut to the chase and use the constructor.
To each their own.
>
>>> somt.getAny().add(0, test);
>> You're not using Java 5.0 that it complains?
> We are using Java 5.
Yes, I gathered later it was a sequence not an feature map...
>
>>> Doing this gives a class cast because it tries to put the 'test'
>>> object into the tkiid field.
>> I would imagine it's trying to add an FeatureMap.Entry to the list...
>>> So clearly I'm not getting how the Sequence
>> Sequence? Is this an SDO thing?
>
> Yes, it is an SDO. The fact that EMF works as SDOs is one of the
> selling points of us using it since it will be compatible with IBMs
> other SDO tooling.
Yes, until they change it to not use EMF. I.e, switch to using SDO 3.0
with something else underneath. Then you can probably start over. :-(
>
>>> coming back from somt.getAny() is supposed to be used.
>>>
>>> All I need to do is insert the object after the tkiid tag when the
>>> object is serialized. Can you offer any example code?
>> This article will give you a better idea of who the FeatureMaps work
>>
>> Binding XML to Java
>>
>> < http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava>
>
> Thanks, I'll look at that and give it a try.
>
>>
>> Personally I think SDO sucks and index-based features as a user API
>> are among the suckiest thing I could imagine. They're totally error
>> prone. Putting them on Sequence is an abomination, given that
>> Sequences are used to deal with wildcards where a feature ID can only
>> be mapped to a feature in the object which can't possibly be the
>> desired thing. There is no magical correct number you can pass to
>> that API for this use case. You'll need to find a Property that
>> corresponds to a global element.
>>
>> I think I'm going to get rid of SDO:
>> <https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402>
>>
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402
>>
>> You'd be better not to use it, in my opinion.
>
> It's probably all 'marketecture' but the SDO badge on EMF makes my
> client accept it.
:-(
> (They're a VERY pro-IBM shop).
IBM has some great technology. EMF being a good example. :-P
> I'm worried that if you remove the SDO stuff they'll either push to
> replace EMF with the SDO stuff from Tuscany
That's a dead end too. I wonder if anyone has noticed.
> or push back on the use of EMF all together. :(
Tell them to get with the program. EMF will be around a lot longer than
anything else they might use today... It's funny how people make what
they think are conservative decisions that ultimately aren't. They
might want to read this, and see how true it rings from their experience.
http://ed-merks.blogspot.com/2008/04/teflon-programming.html
http://ed-merks.blogspot.com/2008/08/engulf-and-devour.html
>
> BTW, I agree with you completely about the index numbers being a
> really bad idea.
Funny how you can't stop a bad idea even when it seems self evidently a
bad idea.
> I just know that I had to refer to EMF object instances as SDOs for
> weeks before they warmed up to the idea of EMF.
You're making me sad.
>
>>>
>>> Thanks,
>>>
>>> Jason Henriksen
--------------030903090702000508000709
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Jason,<br>
<br>
Comments below.<br>
<br>
jason henriksen wrote:
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite">
<blockquote type="cite">
<blockquote type="cite">I try to put an emf object in and I get a
class cast exception:
<br>
<br>
ExecuteHTResponseType test = new ExecuteHTResponseTypeImpl(); &a mp;nbsp;
SetOutputMessageType somt = new SetOutputMessageTypeImpl();
<br>
</blockquote>
That makes me sad. Why aren't you using factories?
<br>
</blockquote>
<br>
They're just verbose. I love factories as a pattern for building code
that may need alteration at runtime. But I can't touch production
without a full recompile/QA/Acceptance flow anyway so why not just say
what I mean? The majority of my objects are business rule
implementations that only make sense in one place anyway. I just make
the objects have public constructors so that I can just get my work
done.
<br>
<br>
If there's any hope that I factory will be useful, I'll leave it. But
if I'm writing sample code, one off rules or other off-the-cuff stuff,
I just cut to the chase and use the constructor.
<br>
</blockquote>
To each their own.<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"><br>
<blockquote type="cite">
<blockquote type="cite"> somt.getAny().add(0, test);
<br>
</blockquote>
You're not using Java 5.0 that it complains?
<br>
</blockquote>
We are using Java 5.
<br>
</blockquote>
Yes, I gathered later it was a sequence not an feature map...<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"><br>
<blockquote type="cite">
<blockquote type="cite">Doing this gives a class cast because it
tries to put the 'test' object into the tkiid field.
<br>
</blockquote>
I would imagine it's trying to add an FeatureMap.Entry to the list...
<br>
<blockquote type="cite">So clearly I'm not getting how the Sequence
</blockquote>
Sequence? Is this an SDO thing?
<br>
</blockquote>
<br>
Yes, it is an SDO. The fact that EMF works as SDOs is one of the
selling points of us using it since it will be compatible with IBMs
other SDO tooling.
<br>
</blockquote>
Yes, until they change it to not use EMF. I.e, switch to using SDO 3.0
with something else underneath. Then you can probably start over. :-(<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"><br>
<blockquote type="cite">
<blockquote type="cite">coming back from somt.getAny() is supposed
to be used.
<br>
<br>
All I need to do is insert the object after the tkiid tag when the
object is serialized. Can you offer any example code?
<br>
</blockquote>
This article will give you a better idea of who the FeatureMaps work
<br>
<br>
Binding XML to Java
<br>
<a class="moz-txt-link-rfc2396E" href=" http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava">< http://www.theserverside.com/tt/articles/article.tss?l=Bindi ngXMLJava></a>
<br>
</blockquote>
<br>
Thanks, I'll look at that and give it a try.
<br>
<br>
<blockquote type="cite"><br>
Personally I think SDO sucks and index-based features as a user API are
among the suckiest thing I could imagine. They're totally error
prone. Putting them on Sequence is an abomination, given that
Sequences are used to deal with wildcards where a feature ID can only
be mapped to a feature in the object which can't possibly be the
desired thing. There is no magical correct number you can pass to
that API for this use case. You'll need to find a Property that
corresponds to a global element.
<br>
<br>
I think I'm going to get rid of SDO:
<br>
<a class="moz-txt-link-rfc2396E" href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402"><https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402></a>
<br>
<br>
<a class="moz-txt-link-freetext" href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402">https://bugs.eclipse.org/bugs/show_bug.cgi?id=251402</a>
<br>
<br>
You'd be better not to use it, in my opinion.
<br>
</blockquote>
<br>
It's probably all 'marketecture' but the SDO badge on EMF makes my
client accept it. <br>
</blockquote>
:-(<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite">
(They're a VERY pro-IBM shop). <br>
</blockquote>
IBM has some great technology. EMF being a good example. :-P<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"> I'm
worried that if you remove the SDO stuff they'll either push to replace
EMF with the SDO stuff from Tuscany</blockquote>
That's a dead end too. I wonder if anyone has noticed.<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"> or
push back on the use of EMF all together. :(
<br>
</blockquote>
Tell them to get with the program. EMF will be around a lot longer
than anything else they might use today... It's funny how people make
what they think are conservative decisions that ultimately aren't.
They might want to read this, and see how true it rings from their
experience.<br>
<blockquote><a
href=" http://ed-merks.blogspot.com/2008/04/teflon-programming.html"> http://ed-merks.blogspot.com/2008/04/teflon-programming.html</a><br>
<a href="http://ed-merks.blogspot.com/2008/08/engulf-and-devour.html">http://ed-merks.blogspot.com/2008/08/engulf-and-devour.html</a><br>
</blockquote>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"><br>
BTW, I agree with you completely about the index numbers being a really
bad idea. <br>
</blockquote>
Funny how you can't stop a bad idea even when it seems self evidently a
bad idea.<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"> I
just know that I had to refer to EMF object instances as SDOs for weeks
before they warmed up to the idea of EMF.
<br>
</blockquote>
You're making me sad.<br>
<blockquote cite="mid:gecvsb$9te$1@build.eclipse.org" type="cite"><br>
<blockquote type="cite">
<blockquote type="cite"><br>
Thanks,
<br>
<br>
Jason Henriksen
<br>
</blockquote>
</blockquote>
</blockquote>
</body>
</html>
--------------030903090702000508000709--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
| |
| Re: xsd:any I'm clearly not using it correctly... [message #424606 is a reply to message #424594] |
Thu, 30 October 2008 21:42 |
Ed Merks
Messages: 33142
Registered: July 2009 |
Senior Member |
|
|
Cameron,
I know all about the FUD. :-P
SDO is a facade that hides EMF, yet typically people drop right down and
use non-SDO things. The use of direct constructors in this thread being
a perfect example.
The SDO specification has gone of in the direction of XML fidelity and
given that no one will look to implement SDO 3.0 around here, it's
realistic for clients to reconsider the long term viability.
Cameron Bateman wrote:
>>> or push back on the use of EMF all together. :(
>> Tell them to get with the program. EMF will be around a lot longer than
> ..
>> might want to read this, and see how true it rings from their
>> experience.
> ..
>> Funny how you can't stop a bad idea even when it seems self evidently a
>
> Or how hard it is to use a good idea that takes more than 5 minutes to
> explain to people. I have no specific opinion on SDO (mainly because
> I have yet to figure out what it's for), but it is troubling to hear
> that something that can be used to sell the use of EMF is going away.
> EMF can be a very hard sell even when it's obvious that it solves a
> specific problem and there are is a lot of FUD around it when you talk
> to certain people. I don't know why, I just know that anything that
> makes my life easier in selling people on using EMF is generally a
> good thing...
>
>
> --Cam
>
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
|
Goto Forum:
Current Time: Fri Apr 26 16:43:01 GMT 2024
Powered by FUDForum. Page generated in 0.03717 seconds |
] 
Search
Help
Register
Login
Home
