Are you familiar with the way we are handling perl modules these days? There is a "perllocal.pod" file which lists properties of all the installed modules. We have the package install this as "perllocal.pod.%n" (roughly), and then in a postinstall script, perllocal.pod is created by concatinating all of the perllocal.pod.%n files together.
Maybe something like this could work for CLASSPATH? Each package providing its ingredients, stored in a good place, and then some file specifying the full CLASSPATH is created in postinstall out of all of these? (Which the user would get by sourcing profile.d ...) Just a rough idea, needs some work, but I think it is feasible. -- Dave Sylvain Cuaz <[EMAIL PROTECTED]> wrote: > Hi, > > In my effort to support Java, I came to 2 conclusions, fink needs to : > 1/ source the profile.d script (if there is one) and, > 2/ recalculate the CLASSPATH (if JarFiles exists) > in postinstScript. This sounds logical to me and would permit the > installation of Java packages on a clean system (1/ would set ANT_HOME > and 2/ would add jars to CLASSPATH as they are installed). > First, does that sound right ? > > Now, the problem is that the postinstScript is executed as root, > so my question is how do I access the user environment ? > > _______________________________________________ Fink-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-devel
