Re: [PROPOSAL] 4. RfD for a new debian java policy

[Ben Burton <bab@debian.org>]

Hi.  Just a couple of notes re the findjava and java-config scripts.

Since the proposed policy mandates that programs use both these scripts
(section 2.6), I would really like to see working implementations of
each of these *before* we look at making this proposal into policy.

>    Packages, which need to find a java virtual machine in the
>    startscript of their programm must use the /usr/bin/findjava
>    programm for this task.

I'm not sure I follow here.  So if I know my app works with kaffe and
gij, and I depend on kaffe and gij, I'm not allowed to simply do
something like:

  for jvm in $JAVA kaffe gij-wrapper; do
    if jvm exists then use it;
  done

but rather I am forced to go through this findjava script?

>    The /usr/bin/java-config programm should be used to setup the
>    classpath.

Again, if my app provides its own JAR and requires nothing else, I'm not
allowed to simply

  export CLASSPATH=/path/to/my/jar

but rather I am forced to go through java-config?

If either of these claims are incorrect, the policy could probably use
some clarification.  Otherwise I would suggest that policy is being a
bit heavy-handed here.

>    Application, which allow to pre-set or overwrite the CLASSPATH
>    need to state this in the manpage and, if possible, at
>    runtime.

What do you mean precisely, state this at runtime?  And what do you mean
"need to" - is this must, should or may?  I personally would propose
this as a "may" - to mandate that programs output what is essentially
debugging information seems a bit much to me.

Ben. :)