7:27 p.m.
Hi,
I have two question about this xsd:
1. why we use xs:string for attributes such as timeout, batch-size, is there a specific
reason?
I'm purposing this change https://gist.github.com/3059390
2. we used to have this with the dtd, but I doubt if it is still possible with in the
current xsd. at least I could not find how to get this query text form
org.hibernate.internal.jaxb.mapping.hbm.JaxbSqlQueryElement
not a xsd guy, can someone take a look?
<sql-query name="orgNamesOnly">
<return-scalar column="NAME" type="string"/>
SELECT org.NAME FROM ORGANIZATION org
</sql-query>
-------------------------
Best Regards,
Strong Liu <stliu at hibernate.org>
http://about.me/stliu/bio
9:44 p.m.
Hi,
first up some info regarding the xsd. We initially auto generated it from our dtd and then
made some adjustments to resolve
some conflicts. The outcome is for sure less than optimal.
I have two question about this xsd:
1. why we use xs:string for attributes such as timeout, batch-size, is there a specific
reason?
I'm purposing this change https://gist.github.com/3059390
I think it is just a result of the transformation. Changing it to xs:int seems reasonable
to me. There are probably
other cases we could improve.
2. we used to have this with the dtd, but I doubt if it is still
possible with in the current xsd. at least I could not find how to get this query text
form org.hibernate.internal.jaxb.mapping.hbm.JaxbSqlQueryElement
not a xsd guy, can someone take a look?
<sql-query name="orgNamesOnly">
<return-scalar column="NAME" type="string"/>
SELECT org.NAME FROM ORGANIZATION org
</sql-query>
What do you get from JaxbSqlQueryElement#getContent. I could imagine that is is part of it
and you need to cast. Probably another spot where we need to do some manual changes.
--Hardy
5:35 p.m.
current <query-element> is defined as :
<xs:complexType name="query-element" mixed="true">
<xs:sequence minOccurs="0" maxOccurs="unbounded">
<xs:element name="query-param"
type="query-param-element"/>
</xs:sequence>
<xs:attribute name="cache-mode"
type="cache-mode-attribute"/>
<xs:attribute name="cache-region" type="xs:string"/>
<xs:attribute name="cacheable" default="false"
type="xs:boolean"/>
<xs:attribute name="comment" type="xs:string"/>
<xs:attribute name="fetch-size" type="xs:int"/>
<xs:attribute name="flush-mode"
type="flush-mode-attribute"/>
<xs:attribute name="name" use="required"
type="xs:string"/>
<xs:attribute name="read-only" type="xs:boolean"/>
<xs:attribute name="timeout" type="xs:int"/>
</xs:complexType>
then the generated org.hibernate.internal.jaxb.mapping.hbm.JaxbQueryElement has a
_getContent()_ method for both String/Text and query-param attribute
* Objects of the following type(s) are allowed in the list
* {@link JAXBElement }{@code <}{@link JaxbQueryParamElement }{@code >}
* {@link String }
not very happy with this, it would be lot better to have a getQueryParam() and getValue()
so we don't need to check the instance type and do the cast everytime.
On Jul 6, 2012, at 8:44 PM, Hardy Ferentschik wrote:
Hi,
first up some info regarding the xsd. We initially auto generated it from our dtd and
then made some adjustments to resolve
some conflicts. The outcome is for sure less than optimal.
> I have two question about this xsd:
>
> 1. why we use xs:string for attributes such as timeout, batch-size, is there a
specific reason?
> I'm purposing this change https://gist.github.com/3059390
I think it is just a result of the transformation. Changing it to xs:int seems reasonable
to me. There are probably
other cases we could improve.
> 2. we used to have this with the dtd, but I doubt if it is still possible with in the
current xsd. at least I could not find how to get this query text form
org.hibernate.internal.jaxb.mapping.hbm.JaxbSqlQueryElement
>
> not a xsd guy, can someone take a look?
>
>
> <sql-query name="orgNamesOnly">
> <return-scalar column="NAME" type="string"/>
> SELECT org.NAME FROM ORGANIZATION org
> </sql-query>
What do you get from JaxbSqlQueryElement#getContent. I could imagine that is is part of
it and you need to cast. Probably another spot where we need to do some manual changes.
--Hardy
-------------------------
Best Regards,
Strong Liu <stliu at hibernate.org>
http://about.me/stliu/bio
5:43 p.m.
Type checking always sucks in this context. I also would try to minimize it as much as
possible.
Provided your suggested change still allows to parse existing mapping documents I
don't see why we should
not make our code easier to write.
--Hardy
On Jul 10, 2012, at 10:35 AM, Strong Liu wrote:
current <query-element> is defined as :
<xs:complexType name="query-element" mixed="true">
<xs:sequence minOccurs="0" maxOccurs="unbounded">
<xs:element name="query-param"
type="query-param-element"/>
</xs:sequence>
<xs:attribute name="cache-mode"
type="cache-mode-attribute"/>
<xs:attribute name="cache-region" type="xs:string"/>
<xs:attribute name="cacheable" default="false"
type="xs:boolean"/>
<xs:attribute name="comment" type="xs:string"/>
<xs:attribute name="fetch-size" type="xs:int"/>
<xs:attribute name="flush-mode"
type="flush-mode-attribute"/>
<xs:attribute name="name" use="required"
type="xs:string"/>
<xs:attribute name="read-only" type="xs:boolean"/>
<xs:attribute name="timeout" type="xs:int"/>
</xs:complexType>
then the generated org.hibernate.internal.jaxb.mapping.hbm.JaxbQueryElement has a
_getContent()_ method for both String/Text and query-param attribute
* Objects of the following type(s) are allowed in the list
* {@link JAXBElement }{@code <}{@link JaxbQueryParamElement }{@code >}
* {@link String }
not very happy with this, it would be lot better to have a getQueryParam() and getValue()
so we don't need to check the instance type and do the cast everytime.
11:04 p.m.
If you can get JAXB to perform that binding for you great! There are
lots of places where we have this.
Just to be clear, however, we cannot change the XSD to accomplish this
in a way that means changes to end-user syntax.
On 07/10/2012 03:43 AM, Hardy Ferentschik wrote:
Type checking always sucks in this context. I also would try to
minimize it as much as possible.
Provided your suggested change still allows to parse existing mapping documents I
don't see why we should
not make our code easier to write.
--Hardy
On Jul 10, 2012, at 10:35 AM, Strong Liu wrote:
> current <query-element> is defined as :
>
> <xs:complexType name="query-element" mixed="true">
> <xs:sequence minOccurs="0" maxOccurs="unbounded">
> <xs:element name="query-param"
type="query-param-element"/>
> </xs:sequence>
> <xs:attribute name="cache-mode"
type="cache-mode-attribute"/>
> <xs:attribute name="cache-region" type="xs:string"/>
> <xs:attribute name="cacheable" default="false"
type="xs:boolean"/>
> <xs:attribute name="comment" type="xs:string"/>
> <xs:attribute name="fetch-size" type="xs:int"/>
> <xs:attribute name="flush-mode"
type="flush-mode-attribute"/>
> <xs:attribute name="name" use="required"
type="xs:string"/>
> <xs:attribute name="read-only" type="xs:boolean"/>
> <xs:attribute name="timeout" type="xs:int"/>
> </xs:complexType>
>
> then the generated org.hibernate.internal.jaxb.mapping.hbm.JaxbQueryElement has a
_getContent()_ method for both String/Text and query-param attribute
>
> * Objects of the following type(s) are allowed in the list
> * {@link JAXBElement }{@code <}{@link JaxbQueryParamElement }{@code >}
> * {@link String }
>
> not very happy with this, it would be lot better to have a getQueryParam() and
getValue() so we don't need to check the instance type and do the cast everytime.
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
--
steve(a)hibernate.org
http://hibernate.org
11:09 p.m.
On Jul 10, 2012, at 4:04 PM, Steve Ebersole wrote:
If you can get JAXB to perform that binding for you great! There are
lots of places where we have this.
Just to be clear, however, we cannot change the XSD to accomplish this in a way that
means changes to end-user syntax.
I think one generic way of doing this would be a custom xjc plugin (these plugins hook
into the code generation and allow to modify
the generated classes in memory before they are written to disc). It might be possible to
add additional type specific getters for mixed content
nodes this way.
I am not saying that we should do it and it is worth the effort, just saying that this
might be a possibility :-)
--Hardy
11:11 p.m.
There are 2 parts though, the class generation and then the binding.
As long as the plugin allows some hook into the binding process as
well...
On Tue 10 Jul 2012 09:09:14 AM CDT, Hardy Ferentschik wrote:
On Jul 10, 2012, at 4:04 PM, Steve Ebersole wrote:
> If you can get JAXB to perform that binding for you great! There are lots of places
where we have this.
>
> Just to be clear, however, we cannot change the XSD to accomplish this in a way that
means changes to end-user syntax.
I think one generic way of doing this would be a custom xjc plugin (these plugins hook
into the code generation and allow to modify
the generated classes in memory before they are written to disc). It might be possible to
add additional type specific getters for mixed content
nodes this way.
I am not saying that we should do it and it is worth the effort, just saying that this
might be a possibility :-)
--Hardy
--
steve(a)hibernate.org
http://hibernate.org
8:21 p.m.
On Jul 10, 2012, at 10:04 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
If you can get JAXB to perform that binding for you great! There are
lots of places where we have this.
yes, I made it :D
https://github.com/stliu/hibernate-orm/commit/d57ffbdb32b8450599f30c9c28a...
https://github.com/stliu/hibernate-orm/commits/HHH-7450
Just to be clear, however, we cannot change the XSD to accomplish this in a way that
means changes to end-user syntax.
On 07/10/2012 03:43 AM, Hardy Ferentschik wrote:
> Type checking always sucks in this context. I also would try to minimize it as much
as possible.
> Provided your suggested change still allows to parse existing mapping documents I
don't see why we should
> not make our code easier to write.
>
> --Hardy
>
> On Jul 10, 2012, at 10:35 AM, Strong Liu wrote:
>
>> current <query-element> is defined as :
>>
>> <xs:complexType name="query-element" mixed="true">
>> <xs:sequence minOccurs="0" maxOccurs="unbounded">
>> <xs:element name="query-param"
type="query-param-element"/>
>> </xs:sequence>
>> <xs:attribute name="cache-mode"
type="cache-mode-attribute"/>
>> <xs:attribute name="cache-region"
type="xs:string"/>
>> <xs:attribute name="cacheable" default="false"
type="xs:boolean"/>
>> <xs:attribute name="comment" type="xs:string"/>
>> <xs:attribute name="fetch-size" type="xs:int"/>
>> <xs:attribute name="flush-mode"
type="flush-mode-attribute"/>
>> <xs:attribute name="name" use="required"
type="xs:string"/>
>> <xs:attribute name="read-only" type="xs:boolean"/>
>> <xs:attribute name="timeout" type="xs:int"/>
>> </xs:complexType>
>>
>> then the generated org.hibernate.internal.jaxb.mapping.hbm.JaxbQueryElement has a
_getContent()_ method for both String/Text and query-param attribute
>>
>> * Objects of the following type(s) are allowed in the list
>> * {@link JAXBElement }{@code <}{@link JaxbQueryParamElement }{@code
>}
>> * {@link String }
>>
>> not very happy with this, it would be lot better to have a getQueryParam() and
getValue() so we don't need to check the instance type and do the cast everytime.
>
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
--
steve(a)hibernate.org
http://hibernate.org
-------------------------
Best Regards,
Strong Liu <stliu at hibernate.org>
http://about.me/stliu/bio
8:32 p.m.
Awesome. Finding the right JAXB customization process/plugin is really like finding the
needle in the haystack.
I will make a great blog entry if we summarize our different techniques.
--Hardy
On Jul 16, 2012, at 1:21 PM, Strong Liu wrote:
On Jul 10, 2012, at 10:04 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
> If you can get JAXB to perform that binding for you great! There are lots of places
where we have this.
yes, I made it :D
https://github.com/stliu/hibernate-orm/commit/d57ffbdb32b8450599f30c9c28a...
https://github.com/stliu/hibernate-orm/commits/HHH-7450
4:11 a.m.
Awesome!
Just make sure that cases where you do this, the ordering amongst the
elements is not important.
On Mon 16 Jul 2012 06:32:37 AM CDT, Hardy Ferentschik wrote:
Awesome. Finding the right JAXB customization process/plugin is
really like finding the needle in the haystack.
I will make a great blog entry if we summarize our different techniques.
--Hardy
On Jul 16, 2012, at 1:21 PM, Strong Liu wrote:
>
> On Jul 10, 2012, at 10:04 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
>
>> If you can get JAXB to perform that binding for you great! There are lots of
places where we have this.
>
> yes, I made it :D
>
>
https://github.com/stliu/hibernate-orm/commit/d57ffbdb32b8450599f30c9c28a...
>
> https://github.com/stliu/hibernate-orm/commits/HHH-7450
>
--
steve(a)hibernate.org
http://hibernate.org
4313
days inactive
4323
days old
9 comments
3 participants
participants (3)
-
Hardy Ferentschik -
Steve Ebersole -
Strong Liu