The reason I'm putting all dependencies in
/src/lib/java (FHS Location)

is that now becomes the central location for all java libraries
/ jar artifacts.

Right now if both Tomcat and ADS are running on the same machine,
they duplicate a lot of the dependencies.

With this there is no duplication.

When OSGi gets going, it also makes
/src/lib/java a "OSGi" service bundle repository,
from which the OSGi container can load all services.

The RPM Factory can insert OSGi manifests.

So if it's hooked up to something like archiva it
can convert an entire repository.  Does depend
on "Clean" pom files though.

The approach also makes download more efficient since
the ADS RPM is minimized.

I think most administrators would prefer the FHS compliant approach.

Anyways - I'm sure others might like the "Uber" RPM as well,
so we could always do both.

Cheers,
- Ole









Chris Custine wrote:
Personally, I favor the self contained, all in one RPM for distribution where we can have better control over dependencies. I wonder if there are any good reasons to do anything else...

Chris

On 5/8/07, *Alex Karasulu* <[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]> > wrote:

    Yeah this is bad.  I figured this would be the case tho since you
    cannot install more than one version of a package and dependencies
    will cause some collisions.  I don't think we can depend on this and
    need to package our dependencies into a single RPM.

    Alex


    On 5/8/07, *Ole Ersoy* < [EMAIL PROTECTED]
    <mailto:[EMAIL PROTECTED]>> wrote:

        I was really excited about JPackage until I found out that they have
        1 version of each "Package" that they support per release.

        Meaning they have JPackage 1.5, 1.6, 1.7...that are "releases"
        of their packages.  Each release should support 1 version of
        1 package.

        Each such release has to have one supported version of a package.

        So suppose Tomcat and ApacheDS share a dependency.

        ApacheDS uses version 1.3 of this dependency in the current
        build.

        Tomcat uses 1.5.

        So what happens?

        Suppose someone at JPackage already built
        version 1.4, and they tried it with both Tomcat and ADS,
        and it looks like it works.

        Well, if that's the supported version in release
        1.x of JPackage, then this dependency could end up getting
        shoehorned
        into a ADS install that JPackage supports.

        Personally I think that's really scary.

        Originally I was writing a Maven plugin for them to automate
        their work, but decided to go in another direction when
        I found out about this policy.

        Cheers,
        - Ole



        Alex Karasulu (JIRA) wrote:
        >     [
        
https://issues.apache.org/jira/browse/DIRSERVER-749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12494305
        
<https://issues.apache.org/jira/browse/DIRSERVER-749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12494305>
        ]
        >
        >  Alex Karasulu commented on DIRSERVER-749:
        >  -----------------------------------------
        >
        >  NP Bastiaan, it's hard to align at times.  Oh I did not know
        about the JPackage rpm.  Perhaps we need to look at that.  None
        of us besides Ole have been in contact with the JPackage
        folks.  Perhaps you can point us in the right direction so we
        can see and discuss what they have done.
        >
        >  BTW Chris Custine is now looking at rewriting some of the code
        in the daemon and installer modules to properly generate an RPM
        with scripts that actually work out of the box.  He's primarily
        focused on the 1.5 branch and will be switching us over to use
        the Tanuki wrapper instead of jsvc and procrun.  As for 1.0 I
        don't think it's worth mucking with.
        >
        > > fix issues with apacheds RPM to get it working out of the box
        > > -------------------------------------------------------------
        > >
        > >                 Key: DIRSERVER-749
        > >                 URL:
        https://issues.apache.org/jira/browse/DIRSERVER-749
        > >             Project: Directory ApacheDS
        > >          Issue Type: Improvement
        > >          Components: installer-plugin
        > >    Affects Versions: 1.0.1, 1.0
        > >         Environment: linux
        > >            Reporter: Bastiaan Bakker
        > >         Assigned To: Alex Karasulu
        > >            Priority: Minor
        > >             Fix For: 1.5.1 , 1.0.3
        > >
        > >         Attachments:
        apacheds-branch-1.0-server-installers-rpmfix.patch,
        apacheds-daemon-trunk-rpmfix.patch
        > >
        > >
        > > The apacheds RPM has several issues that prevent it from
        running out of the box:
        > > * the init script fails to run because APACHEDS_USER is set
        to $USER, which is not defined at boot time
        > > * the init script fails to run bevause JAVA_HOME is not defined
        > > * the init script it is not registered to the init subsystem
        with chkconfig or similar
        > > * the config files are not marked as such, causing them to be
        silently overwritten when one upgrades the RPM
        > > * the RPM filename is not conform conventions:
        ${name}-${version}-${release}.${arch}.rpm
        > > * the location of the files (/usr/local/apacheds-1.0_RC4) is
        version dependent, making upgrades cumbsome. The admin has to
        relocate the partitions and config files on every updgrade.
        > > * the sources and docs are included in the rpm, even though
        they are not necessary for operation.
        > > The RPM build mechanism for apacheds also has some issues:
        > > * runs rpmbuild as root, which is frowned upon by RPM gurus
        for security and safety reasons.
        > > * the generated src.rpm is not self contained, ie. one cannot
        do a 'rpmbuild --rebuild' with it.
        > > * the sudo mechanism is totally unnecessary
        > >
        >



Reply via email to