I've uploaded a preliminary version of the proposed schema jar at
people.apache.org/~djencks/geronimo-schema_1.4_spec-1.0-SNAPSHOT.jar
We should be able to remove all the schemas and xmlbeans stuff from
j2ee-schema module and instead have a geronimo-dependency.xml element
to pull in the above jar.
In m1 these geronimo-dependency.xml files are generated by our
dependency plugin, but for m2 you have to write it yourself and put
it in src/resources2/META-INF/geronimo-dependency.xml
I have some more work to do on the proposed new module to put the
xmlbeans stuff in a separate profile so the source is not generated
each time it's run. I'll open a jira with the new module when I get
it working better. I assume this has to go through RTC.... maybe it
can be kind of quick since we are basically just changing where we
build and compile the exact same classes as before.
thanks
david jencks
On Aug 18, 2006, at 10:46 AM, David Jencks wrote:
On Aug 18, 2006, at 10:36 AM, David Blevins wrote:
I personally don't know why we bother to regenerate that tree on
every build as those schemas are not going to change. Let's just
build them once, publish the jars, delete the schemas and be done
with it.
I think this will minimize our legal exposure. I'll work on a
specs module for this. I might need some m2 help, my plan is to
use a profile so that using the profile we compile the schemas
using xmlbeans, letting xmlbeans download the schemas direct from
sun, and having xmlbeans put the results in appropriate places in
src. We can then check the results in by hand. The normal build
can simply compile these generated sources.
We may need another step to put apache headers on the generated
code. Some of the generated stuff is binary files, which AFAIK
cannot be modified to include a header. I'll probably need help
with this part.
Is everyone OK with putting this into specs? Any ideas for a
better place?
thanks
david jencks
-David
On Aug 18, 2006, at 9:45 AM, David Jencks wrote:
On Aug 18, 2006, at 9:22 AM, Geir Magnusson Jr wrote:
If they are only used during the build process, then don't re-
distribute
and therefore this issue then seems out of the critical path for
1.1.1
and can be solved for 1.1.1.1.1.1.1.1.1
They are not needed at runtime, but xmlbeans packs the source
schemas into the jar with its other binary artifacts. We might
be able to prevent their inclusion into our jars, but I don't see
how having them in svn is not redistribution, so I think we need
allowable copies of the j2ee 1.4 schemas in our svn.
thanks
david jencks
gier
Matt Hogstrom wrote:
They are used by XMLBeans during the build process. I guess we
could do
a one-time generation of the XMLBeans classes but I'd be more
comfortable with Jencks' input. If you don't have the schemas
you'd
have to have network connectivity to build I suspect.
Aaron Mulder wrote:
Can't we just ship without those? We have pointers to them on
our web
site at http://geronimo.apache.org/schemas.html -- is there any
additional reason we need them at runtime (e.g. does XMLBeans
use the
actual schema to validate at runtime)?
Thanks,
Aaron
On 8/18/06, Matt Hogstrom <[EMAIL PROTECTED]> wrote:
All,
For those wondering where the Geronimo 1.1.1 release is at
here is a
quick summary and battle plan.
John Sisson discovered that we have several DTD and XSDs
included in
our build that are copies of
Sun's original material. The copyright in the material seems to
indicate that we cannot
redistribute the content. Unfortunately, as Geir has pointed
out,
that there is no clear way to
interpret the license and we should manually generate these
to stay
legal and avoid any appearance
of impropriety.
The documents in question reside in both the distribution of the
source as well as the server
itself. It appears that other open source and commercial
vendors do
distribute these documents in
their original form but the ASF does not have a clear
indication that
we can follow this same
practice. You can find the documents in
$G_BUILD_TREE/modules/j2ee-schema/src/* as well as the
distributions in $G_DIST/schema/* directories.
The documents are identified in JIRA
http://issues.apache.org/jira/browse/GERONIMO-2307.
I'd like to ask anyone that has some spare moments to help
out with
putting these together.
Basically, you need to refer to the appropriate specification
and
type in the XML exactly like it is
represented for actual elements and either omit or paraphrase
the
information embedded in comments
so we do not violate Sun's copyright.
I know this is frustrating but for their own reasons Sun has
imposed
this unfriendly copyright and
we need to abide by it. We are protecting The ASF, Geronimo
and our
users.
If you are working on a document please update the above JIRA to
indicate it is partial so others
can see what remains. Please check the new schemas into their
respective replacements. When the
replacements are complete we'll build a new distribution and
run it
through a full CTS test to
validate our work.
Please also consider working with a partner so you can cross
check
each other's work.
Thanks in advance for your patience and help with this
important issue.
Also, if others have input into the process that I have
missed please
provide it as well.
Thanks to John and Geir for discovering and mediating on this
issue.
