AFAIK there is a way in maven repo to "relocate" dependencies, so that a
POM for any commons can be published at commons-xxx groupId, that
"relocates" to org.apache.commons" groupId.
Servletapi for example is now under "javax.servlet"
http://www.ibiblio.org/maven2/servletapi/servletapi/2.4/servletapi-2.4.pom
Using this, when maven2 search for the "latest" release of any commons
it will look at the relocated one.
Torsten Curdt a écrit :
Brett,
any comments on this?
cheers
--
Torsten
On 6/6/06, Dennis Lundberg <[EMAIL PROTECTED]> wrote:
Brett, I did the test that you suggested.
1. Installed commons-lang 1.0.1 into my local repo with
groupId=org.apache.commons
mvn install:install-file -DgroupId=org.apache.commons
-DartifactId=commons-lang -Dversion=1.0.1 -Dpackaging=jar
-Dfile=/path/to/commons-lang-1.0.1.jar
2. Created Maven 2 projects a, b and c with the dependencies mentioned
below.
3. Installed projects a and b into my local repo
mvn install
4. packaged project c as a war
mvn package
The resulting war file includes both commons-lang-1.0.1.jar and
commons-lang-2.1.jar which was what you thought would happen.
So this is bad, I guess. Anyone who uses commons components transitively
in a Maven 2 environment are likely to be bitten by this. They must keep
the same groupId for all commons-lang dependencies, as an example, in
the entire chain of transitive dependencies. I.e. they can't mix
groupId=commons-lang and groupId=org.apache.commons. This can be a PITA
since some of the dependencies are most likely out of the projects own
control.
What do you suggest we do? Should we wait with this relocation until a
version of Maven 2 is released that can handle these kind of
dependencies?
--
Dennis Lundberg
Brett Porter wrote:
> an extensive test should be something along the lines of:
>
> project A depends on commons-lang:commons-lang 2.1
> project B depends on o.a.c:commons-lang 1.0
> project C is a webapp that depends on A and B
>
> webapp should have only one commons-lang.
>
> You could do this with your own repository (and something completely
> artificial instead of commons-lang if it makes it easier).
>
> - Brett
>
> Dennis Lundberg wrote:
>> Hi Brett
>>
>> Sorry, I misunderstood you regarding when to do the testing. So, no I
>> haven't done the test, yet. Can you elaborate a bit more on what
needs
>> to be tested? Perhaps you know of an artifact that has been relocated
>> that we can have a look at, to see how they have done.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
This message contains information that may be privileged or confidential and is
the property of the Capgemini Group. It is intended only for the person to whom
it is addressed. If you are not the intended recipient, you are not authorized
to read, print, retain, copy, disseminate, distribute, or use this message or
any part thereof. If you receive this message in error, please notify the
sender immediately and delete all copies of this message.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
