You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jdo-dev@db.apache.org by Craig Russell <Cr...@sun.com> on 2005/03/20 00:25:57 UTC
DTD files
Team,
As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I had
to think again about where we should put the official PUBLIC dtd files.
For JDO 1, it was clear that the site should be Sun. For JDO 2, I
believe that the answer still should be Sun, because the JCP officially
owns the IP to the dtd (and xml schema once we have that in place).
Does anyone think that we should change the official web location to
some other place like the Apache web site?
Brian,
The dtd files I checked in reflect the latest JDO specification draft.
I understand you're interested in converting these to .xsd. The last
time I saw an xsd for jdo was when Robin Roos made one for an earlier
draft. He might be able to help get you started with the task.
Craig
Craig Russell
Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
408 276-5638 mailto:Craig.Russell@sun.com
P.S. A good JDO? O, Gasp!
Re: DTD files
Posted by Brian Topping <to...@codehaus.org>.
Craig Russell wrote:
> So far, we have not had a need to make the jdo and orm doctypes more
> expressive than we can accomplish using DTD. In other words, there's
> no technical reason to drive toward xsd.
>
I guess I should also add that I wrote a tool for Dentaku a few months
back that parses XSD and creates tagged values in XMI for the schema.
This makes MDA very easy. http://www.dentaku.org/Cartridge+Generator
describes this tool if anyone is interested.
Thus, the motivation behind having XSD as the normative spec is that any
tool like this tends to expose weaknesses in the metadata for projects,
and having to correct a DTD in a way that the machine translation of it
will be correct is a bit obtuse. Is this something that we care about?
If not, that's okay, but developer automation is inevitable and DTD is
obsolete in large part because of this. (Arguably, DTD has hung around
because tools like this haven't materialized as quickly as originally
thought.) MOF 2.0 is already oriented toward generating XSD as a
standard artifact, for instance.
$0.02 more....
-b
Re: DTD files
Posted by Geir Magnusson Jr <ge...@4quarters.com>.
On Mar 21, 2005, at 3:08 PM, Brian McCallister wrote:
> I am in favor of XSD just because of all the tool support for it.
> RelaxNG doesn't have that, yet.
>
I'm in favor of DTD because of all the tool support you need for XSD :)
geir
> -Brian
>
> On Mar 21, 2005, at 2:46 PM, Matthew T. Adams wrote:
>
>> I would like to see an XSD. I don't think that it's that hard to
>> read,
>> plus the tools out there are really good nowadays to provide
>> completion
>> of elements and attributes.
>>
>> I'd like to stay away from RelaxNG, only because I don't see the XSD
>> as
>> being too hard.
>>
>> Just my $0.02.
>>
>> --matthew
>>
>>> -----Original Message-----
>>> From: Craig Russell [mailto:Craig.Russell@Sun.COM]
>>> Sent: Monday, March 21, 2005 11:34 AM
>>> To: jdo-dev@db.apache.org
>>> Subject: Re: DTD files
>>>
>>>
>>> Hi Brian,
>>>
>>> We have discussed in the JDO expert group that the standard will be
>>> published as xsd, but I haven't had time to make the change in
>>> the spec
>>> itself. The idea I had (which might be completely whack) is to
>>> publish
>>> the xsd as the norm, and the DTD as non-normative. It is much
>>> easier to
>>> read a tight DTD than the equivalent xsd IMHO.
>>>
>>> So far, we have not had a need to make the jdo and orm doctypes more
>>> expressive than we can accomplish using DTD. In other words,
>>> there's no
>>> technical reason to drive toward xsd.
>>>
>>> If RelaxNG is suitable for use in Apache, I'd entertain using it for
>>> our purposes. Do you have details on its closure?
>>>
>>> Craig
>>>
>>> On Mar 21, 2005, at 11:22 AM, Brian Topping wrote:
>>>
>>>> Hi Craig,
>>>>
>>>> Yes, this is good. I'll take a look at that. From the conference
>>>> call week back (sorry I missed everyone this week), we were
>>>> considering that the public metadata specs ought to be in
>>> XSD, without
>>>> DTD. What do you think of that? I'm not a big DTD user, so I don't
>>>> know what the reasons are that people commonly want to use
>>> them. The
>>>> advantages of using XSD are that there is only one spec, that XSD is
>>>> much more expressive, and it can be parsed with standard XML tools.
>>>> If we are still going to maintain a DTD, I would suggest
>>> that it is a
>>>> mistake to develop an XSD that is not a machine translation of the
>>>> DTD, because no XSD is a better answer than an incorrect XSD
>>> resulting
>>>> from not being maintained. And if we are going to do a machine
>>>> translation, I'd just suggest putting Trang into the build
>>> and make it
>>>> a build-time artifact rather than checking it it. It's a pretty
>>>> slippery slope if we don't want to stay focused on XSD.
>>>>
>>>> Another alternate is to use RelaxNG, which is reasonably expressive,
>>>> can be written in XML (so it is can be parsed), and can be machine
>>>> translated into both DTD and XSD. It is not as expressive
>>> as XSD, but
>>>> if I remember correctly, there are only a couple of XSD
>>> constructions
>>>> that can't be done in it. It does add the RelaxNG closure to the
>>>> dependencies though.
>>>>
>>>> Comments anyone?
>>>>
>>>> -b
>>>>
>>>> Craig Russell wrote:
>>>>
>>>>> Team,
>>>>>
>>>>> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
>>>>> had to think again about where we should put the official
>>> PUBLIC dtd
>>>>> files. For JDO 1, it was clear that the site should be Sun. For JDO
>>>>> 2, I believe that the answer still should be Sun, because the JCP
>>>>> officially owns the IP to the dtd (and xml schema once we have that
>>>>> in place).
>>>>>
>>>>> Does anyone think that we should change the official web
>>> location to
>>>>> some other place like the Apache web site?
>>>>>
>>>>> Brian,
>>>>>
>>>>> The dtd files I checked in reflect the latest JDO specification
>>>>> draft. I understand you're interested in converting these to .xsd.
>>>>> The last time I saw an xsd for jdo was when Robin Roos made one for
>>>>> an earlier draft. He might be able to help get you started with the
>>>>> task.
>>>>>
>>>>> Craig
>>>>>
>>>>> Craig Russell
>>>>> Architect, Sun Java Enterprise System
>>> http://java.sun.com/products/jdo
>>>>> 408 276-5638
>>> mailto:Craig.Russell@sun.com
>>>>> P.S. A good JDO? O, Gasp!
>>>>
>>>>
>>> Craig Russell
>>> Architect, Sun Java Enterprise System
>>> http://java.sun.com/products/jdo
>>> 408 276-5638 mailto:Craig.Russell@sun.com
>>> P.S. A good JDO? O, Gasp!
>>>
>>
>>
>>
>
>
--
Geir Magnusson Jr +1-203-665-6437
geirm@apache.org
Re: DTD files
Posted by Brian Topping <to...@codehaus.org>.
I wanted to present the options as I saw them in an unbiased manner, but
I must admit to being in agreement with Brian and Matthew. I don't have
anything against DTD, but when compared to XSD, I don't prefer it. It
can be more complicated, but after working with it for some time now, I
actually find it a bit easier to understand. Familiarity, I guess.
The point about tool support and "control-space" style typing completion
in tools is a big thing. With this, you can write a well-formed and
valid document without even opening the DTD or Schema. In IDEA Irida
(the current development version, which is quite stable), you can also
use their "quick JavaDoc" command to bring up a tooltip-like popup with
documentation for a node. In that regard, the experience is the same,
whether you use DTD or schema.
The biggest issue for me is maintenance of the non-normative standard.
I think this ought to be a build-time artifact with a tool, not
something that the maintainers are required to maintain. A disadvantage
of XSD is that it is not commonly translatable into other standards, but
that is because it is semantically richer and there are a few
constructions that can't be translated. Hence, I would prefer to drop
DTD altogether, but it's true that there shouldn't *be* maintenance of
the metadata after the standard is finished, so this is arguably a moot
point.
-b
Brian McCallister wrote:
> I am in favor of XSD just because of all the tool support for it.
> RelaxNG doesn't have that, yet.
>
> -Brian
>
> On Mar 21, 2005, at 2:46 PM, Matthew T. Adams wrote:
>
>> I would like to see an XSD. I don't think that it's that hard to read,
>> plus the tools out there are really good nowadays to provide completion
>> of elements and attributes.
>>
>> I'd like to stay away from RelaxNG, only because I don't see the XSD as
>> being too hard.
>>
>> Just my $0.02.
>>
>> --matthew
>>
>>> -----Original Message-----
>>> From: Craig Russell [mailto:Craig.Russell@Sun.COM]
>>> Sent: Monday, March 21, 2005 11:34 AM
>>> To: jdo-dev@db.apache.org
>>> Subject: Re: DTD files
>>>
>>>
>>> Hi Brian,
>>>
>>> We have discussed in the JDO expert group that the standard will be
>>> published as xsd, but I haven't had time to make the change in
>>> the spec
>>> itself. The idea I had (which might be completely whack) is to publish
>>> the xsd as the norm, and the DTD as non-normative. It is much
>>> easier to
>>> read a tight DTD than the equivalent xsd IMHO.
>>>
>>> So far, we have not had a need to make the jdo and orm doctypes more
>>> expressive than we can accomplish using DTD. In other words,
>>> there's no
>>> technical reason to drive toward xsd.
>>>
>>> If RelaxNG is suitable for use in Apache, I'd entertain using it for
>>> our purposes. Do you have details on its closure?
>>>
>>> Craig
>>>
>>> On Mar 21, 2005, at 11:22 AM, Brian Topping wrote:
>>>
>>>> Hi Craig,
>>>>
>>>> Yes, this is good. I'll take a look at that. From the conference
>>>> call week back (sorry I missed everyone this week), we were
>>>> considering that the public metadata specs ought to be in
>>>
>>> XSD, without
>>>
>>>> DTD. What do you think of that? I'm not a big DTD user, so I don't
>>>> know what the reasons are that people commonly want to use
>>>
>>> them. The
>>>
>>>> advantages of using XSD are that there is only one spec, that XSD is
>>>> much more expressive, and it can be parsed with standard XML tools.
>>>> If we are still going to maintain a DTD, I would suggest
>>>
>>> that it is a
>>>
>>>> mistake to develop an XSD that is not a machine translation of the
>>>> DTD, because no XSD is a better answer than an incorrect XSD
>>>
>>> resulting
>>>
>>>> from not being maintained. And if we are going to do a machine
>>>> translation, I'd just suggest putting Trang into the build
>>>
>>> and make it
>>>
>>>> a build-time artifact rather than checking it it. It's a pretty
>>>> slippery slope if we don't want to stay focused on XSD.
>>>>
>>>> Another alternate is to use RelaxNG, which is reasonably expressive,
>>>> can be written in XML (so it is can be parsed), and can be machine
>>>> translated into both DTD and XSD. It is not as expressive
>>>
>>> as XSD, but
>>>
>>>> if I remember correctly, there are only a couple of XSD
>>>
>>> constructions
>>>
>>>> that can't be done in it. It does add the RelaxNG closure to the
>>>> dependencies though.
>>>>
>>>> Comments anyone?
>>>>
>>>> -b
>>>>
>>>> Craig Russell wrote:
>>>>
>>>>> Team,
>>>>>
>>>>> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
>>>>> had to think again about where we should put the official
>>>>
>>> PUBLIC dtd
>>>
>>>>> files. For JDO 1, it was clear that the site should be Sun. For JDO
>>>>> 2, I believe that the answer still should be Sun, because the JCP
>>>>> officially owns the IP to the dtd (and xml schema once we have that
>>>>> in place).
>>>>>
>>>>> Does anyone think that we should change the official web
>>>>
>>> location to
>>>
>>>>> some other place like the Apache web site?
>>>>>
>>>>> Brian,
>>>>>
>>>>> The dtd files I checked in reflect the latest JDO specification
>>>>> draft. I understand you're interested in converting these to .xsd.
>>>>> The last time I saw an xsd for jdo was when Robin Roos made one for
>>>>> an earlier draft. He might be able to help get you started with the
>>>>> task.
>>>>>
>>>>> Craig
>>>>>
>>>>> Craig Russell
>>>>> Architect, Sun Java Enterprise System
>>>>
>>> http://java.sun.com/products/jdo
>>>
>>>>> 408 276-5638
>>>>
>>> mailto:Craig.Russell@sun.com
>>>
>>>>> P.S. A good JDO? O, Gasp!
>>>>
>>>>
>>>>
>>> Craig Russell
>>> Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
>>> 408 276-5638 mailto:Craig.Russell@sun.com
>>> P.S. A good JDO? O, Gasp!
>>>
>>
>>
>>
>
>
Re: DTD files
Posted by Brian McCallister <br...@apache.org>.
I am in favor of XSD just because of all the tool support for it.
RelaxNG doesn't have that, yet.
-Brian
On Mar 21, 2005, at 2:46 PM, Matthew T. Adams wrote:
> I would like to see an XSD. I don't think that it's that hard to read,
> plus the tools out there are really good nowadays to provide completion
> of elements and attributes.
>
> I'd like to stay away from RelaxNG, only because I don't see the XSD as
> being too hard.
>
> Just my $0.02.
>
> --matthew
>
>> -----Original Message-----
>> From: Craig Russell [mailto:Craig.Russell@Sun.COM]
>> Sent: Monday, March 21, 2005 11:34 AM
>> To: jdo-dev@db.apache.org
>> Subject: Re: DTD files
>>
>>
>> Hi Brian,
>>
>> We have discussed in the JDO expert group that the standard will be
>> published as xsd, but I haven't had time to make the change in
>> the spec
>> itself. The idea I had (which might be completely whack) is to publish
>> the xsd as the norm, and the DTD as non-normative. It is much
>> easier to
>> read a tight DTD than the equivalent xsd IMHO.
>>
>> So far, we have not had a need to make the jdo and orm doctypes more
>> expressive than we can accomplish using DTD. In other words,
>> there's no
>> technical reason to drive toward xsd.
>>
>> If RelaxNG is suitable for use in Apache, I'd entertain using it for
>> our purposes. Do you have details on its closure?
>>
>> Craig
>>
>> On Mar 21, 2005, at 11:22 AM, Brian Topping wrote:
>>
>>> Hi Craig,
>>>
>>> Yes, this is good. I'll take a look at that. From the conference
>>> call week back (sorry I missed everyone this week), we were
>>> considering that the public metadata specs ought to be in
>> XSD, without
>>> DTD. What do you think of that? I'm not a big DTD user, so I don't
>>> know what the reasons are that people commonly want to use
>> them. The
>>> advantages of using XSD are that there is only one spec, that XSD is
>>> much more expressive, and it can be parsed with standard XML tools.
>>> If we are still going to maintain a DTD, I would suggest
>> that it is a
>>> mistake to develop an XSD that is not a machine translation of the
>>> DTD, because no XSD is a better answer than an incorrect XSD
>> resulting
>>> from not being maintained. And if we are going to do a machine
>>> translation, I'd just suggest putting Trang into the build
>> and make it
>>> a build-time artifact rather than checking it it. It's a pretty
>>> slippery slope if we don't want to stay focused on XSD.
>>>
>>> Another alternate is to use RelaxNG, which is reasonably expressive,
>>> can be written in XML (so it is can be parsed), and can be machine
>>> translated into both DTD and XSD. It is not as expressive
>> as XSD, but
>>> if I remember correctly, there are only a couple of XSD
>> constructions
>>> that can't be done in it. It does add the RelaxNG closure to the
>>> dependencies though.
>>>
>>> Comments anyone?
>>>
>>> -b
>>>
>>> Craig Russell wrote:
>>>
>>>> Team,
>>>>
>>>> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
>>>> had to think again about where we should put the official
>> PUBLIC dtd
>>>> files. For JDO 1, it was clear that the site should be Sun. For JDO
>>>> 2, I believe that the answer still should be Sun, because the JCP
>>>> officially owns the IP to the dtd (and xml schema once we have that
>>>> in place).
>>>>
>>>> Does anyone think that we should change the official web
>> location to
>>>> some other place like the Apache web site?
>>>>
>>>> Brian,
>>>>
>>>> The dtd files I checked in reflect the latest JDO specification
>>>> draft. I understand you're interested in converting these to .xsd.
>>>> The last time I saw an xsd for jdo was when Robin Roos made one for
>>>> an earlier draft. He might be able to help get you started with the
>>>> task.
>>>>
>>>> Craig
>>>>
>>>> Craig Russell
>>>> Architect, Sun Java Enterprise System
>> http://java.sun.com/products/jdo
>>>> 408 276-5638
>> mailto:Craig.Russell@sun.com
>>>> P.S. A good JDO? O, Gasp!
>>>
>>>
>> Craig Russell
>> Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
>> 408 276-5638 mailto:Craig.Russell@sun.com
>> P.S. A good JDO? O, Gasp!
>>
>
>
>
RE: DTD files
Posted by "Matthew T. Adams" <ma...@xcalia.com>.
I would like to see an XSD. I don't think that it's that hard to read,
plus the tools out there are really good nowadays to provide completion
of elements and attributes.
I'd like to stay away from RelaxNG, only because I don't see the XSD as
being too hard.
Just my $0.02.
--matthew
>-----Original Message-----
>From: Craig Russell [mailto:Craig.Russell@Sun.COM]
>Sent: Monday, March 21, 2005 11:34 AM
>To: jdo-dev@db.apache.org
>Subject: Re: DTD files
>
>
>Hi Brian,
>
>We have discussed in the JDO expert group that the standard will be
>published as xsd, but I haven't had time to make the change in
>the spec
>itself. The idea I had (which might be completely whack) is to publish
>the xsd as the norm, and the DTD as non-normative. It is much
>easier to
>read a tight DTD than the equivalent xsd IMHO.
>
>So far, we have not had a need to make the jdo and orm doctypes more
>expressive than we can accomplish using DTD. In other words,
>there's no
>technical reason to drive toward xsd.
>
>If RelaxNG is suitable for use in Apache, I'd entertain using it for
>our purposes. Do you have details on its closure?
>
>Craig
>
>On Mar 21, 2005, at 11:22 AM, Brian Topping wrote:
>
>> Hi Craig,
>>
>> Yes, this is good. I'll take a look at that. From the conference
>> call week back (sorry I missed everyone this week), we were
>> considering that the public metadata specs ought to be in
>XSD, without
>> DTD. What do you think of that? I'm not a big DTD user, so I don't
>> know what the reasons are that people commonly want to use
>them. The
>> advantages of using XSD are that there is only one spec, that XSD is
>> much more expressive, and it can be parsed with standard XML tools.
>> If we are still going to maintain a DTD, I would suggest
>that it is a
>> mistake to develop an XSD that is not a machine translation of the
>> DTD, because no XSD is a better answer than an incorrect XSD
>resulting
>> from not being maintained. And if we are going to do a machine
>> translation, I'd just suggest putting Trang into the build
>and make it
>> a build-time artifact rather than checking it it. It's a pretty
>> slippery slope if we don't want to stay focused on XSD.
>>
>> Another alternate is to use RelaxNG, which is reasonably expressive,
>> can be written in XML (so it is can be parsed), and can be machine
>> translated into both DTD and XSD. It is not as expressive
>as XSD, but
>> if I remember correctly, there are only a couple of XSD
>constructions
>> that can't be done in it. It does add the RelaxNG closure to the
>> dependencies though.
>>
>> Comments anyone?
>>
>> -b
>>
>> Craig Russell wrote:
>>
>>> Team,
>>>
>>> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
>>> had to think again about where we should put the official
>PUBLIC dtd
>>> files. For JDO 1, it was clear that the site should be Sun. For JDO
>>> 2, I believe that the answer still should be Sun, because the JCP
>>> officially owns the IP to the dtd (and xml schema once we have that
>>> in place).
>>>
>>> Does anyone think that we should change the official web
>location to
>>> some other place like the Apache web site?
>>>
>>> Brian,
>>>
>>> The dtd files I checked in reflect the latest JDO specification
>>> draft. I understand you're interested in converting these to .xsd.
>>> The last time I saw an xsd for jdo was when Robin Roos made one for
>>> an earlier draft. He might be able to help get you started with the
>>> task.
>>>
>>> Craig
>>>
>>> Craig Russell
>>> Architect, Sun Java Enterprise System
>http://java.sun.com/products/jdo
>>> 408 276-5638
>mailto:Craig.Russell@sun.com
>>> P.S. A good JDO? O, Gasp!
>>
>>
>Craig Russell
>Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
>408 276-5638 mailto:Craig.Russell@sun.com
>P.S. A good JDO? O, Gasp!
>
Re: DTD files
Posted by Craig Russell <Cr...@Sun.COM>.
Hi Brian,
We have discussed in the JDO expert group that the standard will be
published as xsd, but I haven't had time to make the change in the spec
itself. The idea I had (which might be completely whack) is to publish
the xsd as the norm, and the DTD as non-normative. It is much easier to
read a tight DTD than the equivalent xsd IMHO.
So far, we have not had a need to make the jdo and orm doctypes more
expressive than we can accomplish using DTD. In other words, there's no
technical reason to drive toward xsd.
If RelaxNG is suitable for use in Apache, I'd entertain using it for
our purposes. Do you have details on its closure?
Craig
On Mar 21, 2005, at 11:22 AM, Brian Topping wrote:
> Hi Craig,
>
> Yes, this is good. I'll take a look at that. From the conference
> call week back (sorry I missed everyone this week), we were
> considering that the public metadata specs ought to be in XSD, without
> DTD. What do you think of that? I'm not a big DTD user, so I don't
> know what the reasons are that people commonly want to use them. The
> advantages of using XSD are that there is only one spec, that XSD is
> much more expressive, and it can be parsed with standard XML tools.
> If we are still going to maintain a DTD, I would suggest that it is a
> mistake to develop an XSD that is not a machine translation of the
> DTD, because no XSD is a better answer than an incorrect XSD resulting
> from not being maintained. And if we are going to do a machine
> translation, I'd just suggest putting Trang into the build and make it
> a build-time artifact rather than checking it it. It's a pretty
> slippery slope if we don't want to stay focused on XSD.
>
> Another alternate is to use RelaxNG, which is reasonably expressive,
> can be written in XML (so it is can be parsed), and can be machine
> translated into both DTD and XSD. It is not as expressive as XSD, but
> if I remember correctly, there are only a couple of XSD constructions
> that can't be done in it. It does add the RelaxNG closure to the
> dependencies though.
>
> Comments anyone?
>
> -b
>
> Craig Russell wrote:
>
>> Team,
>>
>> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
>> had to think again about where we should put the official PUBLIC dtd
>> files. For JDO 1, it was clear that the site should be Sun. For JDO
>> 2, I believe that the answer still should be Sun, because the JCP
>> officially owns the IP to the dtd (and xml schema once we have that
>> in place).
>>
>> Does anyone think that we should change the official web location to
>> some other place like the Apache web site?
>>
>> Brian,
>>
>> The dtd files I checked in reflect the latest JDO specification
>> draft. I understand you're interested in converting these to .xsd.
>> The last time I saw an xsd for jdo was when Robin Roos made one for
>> an earlier draft. He might be able to help get you started with the
>> task.
>>
>> Craig
>>
>> Craig Russell
>> Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
>> 408 276-5638 mailto:Craig.Russell@sun.com
>> P.S. A good JDO? O, Gasp!
>
>
Craig Russell
Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
408 276-5638 mailto:Craig.Russell@sun.com
P.S. A good JDO? O, Gasp!
Re: DTD files
Posted by Brian Topping <to...@codehaus.org>.
Hi Craig,
Yes, this is good. I'll take a look at that. From the conference call
week back (sorry I missed everyone this week), we were considering that
the public metadata specs ought to be in XSD, without DTD. What do you
think of that? I'm not a big DTD user, so I don't know what the reasons
are that people commonly want to use them. The advantages of using XSD
are that there is only one spec, that XSD is much more expressive, and
it can be parsed with standard XML tools.
If we are still going to maintain a DTD, I would suggest that it is a
mistake to develop an XSD that is not a machine translation of the DTD,
because no XSD is a better answer than an incorrect XSD resulting from
not being maintained. And if we are going to do a machine translation,
I'd just suggest putting Trang into the build and make it a build-time
artifact rather than checking it it. It's a pretty slippery slope if we
don't want to stay focused on XSD.
Another alternate is to use RelaxNG, which is reasonably expressive, can
be written in XML (so it is can be parsed), and can be machine
translated into both DTD and XSD. It is not as expressive as XSD, but
if I remember correctly, there are only a couple of XSD constructions
that can't be done in it. It does add the RelaxNG closure to the
dependencies though.
Comments anyone?
-b
Craig Russell wrote:
> Team,
>
> As I was checking in the dtd files for jdo2 (jdo.dtd and orm.dtd) I
> had to think again about where we should put the official PUBLIC dtd
> files. For JDO 1, it was clear that the site should be Sun. For JDO 2,
> I believe that the answer still should be Sun, because the JCP
> officially owns the IP to the dtd (and xml schema once we have that in
> place).
>
> Does anyone think that we should change the official web location to
> some other place like the Apache web site?
>
> Brian,
>
> The dtd files I checked in reflect the latest JDO specification draft.
> I understand you're interested in converting these to .xsd. The last
> time I saw an xsd for jdo was when Robin Roos made one for an earlier
> draft. He might be able to help get you started with the task.
>
> Craig
>
> Craig Russell
> Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
> 408 276-5638 mailto:Craig.Russell@sun.com
> P.S. A good JDO? O, Gasp!