On Thu, 26 Mar 2026 at 12:13, Rob Tompkins <[email protected]> wrote:
>
>
>
> > On Mar 26, 2026, at 7:55 AM, sebb <[email protected]> wrote:
> >
> > On Thu, 26 Mar 2026 at 11:37, Elliotte Rusty Harold <[email protected]> 
> > wrote:
> >>
> >> On Wed, Mar 25, 2026 at 7:58 PM Thomas Vandahl <[email protected]> wrote:
> >>>
> >>> Hi folks,
> >>>
> >>> I would like to build the site for JCS 4.0.0. The japicmp report wants to 
> >>> compare the current jar with the previous one. In commons-parent, the 
> >>> configuration uses <artifactId>${project.artifactId}</artifactId> with 
> >>> the <version>${commons.bc.version}</version>.
> >>>
> >>> Now, as the rule for major releases is to change the artifactId (from 
> >>> commons-jcs3-core to commons-jcs4-core), this reference point to 
> >>> commons-jcs4-core:3.2.1, which of course doesn't exist.
> >>
> >> Did the Java package name also change? They need to be synced with the
> >> artifact ID.
> >
> > Agreed.
> >
> > But does japicmp really apply to major version changes?
> > These will generally involve breaking API changes.
>
> Perhaps it’s nice to see for the purposes of version migration from a user 
> perspective. But, from a validation perspective your point is correct the 
> report provides moot evidence.
>
> >
> > It should be possible to override the previous version of the jar, but
> > this will presumably report everything as a difference, which would
> > not be particularly useful.
> >
> > I don't think japicmp has an option to ignore package name changes, so
> > one of the jars would need to be shaded so the two jars used the same
> > package names.
> >
> > I don't know how easy this would be to do in Maven; probably best to
> > try manually first.
>
> +/- 0 on inclusion for major version changes. May help with forward adoption, 
> but also may be confusing and klugey (unclear to me which is better). If we 
> were to use it to write migration path documentation then set aside the 
> japicmp and kept the documentation that would be ideal, but alas forward 
> progress needs to be mindful of project workload/capacity. I know not what to 
> do.

AFACT japicmp has a standalone version, and I have just found what is
billed as a standalone version of maven shading.
This looks easy to use:
https://github.com/lucko/jar-relocator

If no-one beats me to it, I might try it ...

Sebb
> Good ideas, good conversation!
>
> Thank you Elliotte, Sebb, Thomas!
>
> Cheers,
> -Rob
>
> >
> > Sebb
> >
> >> --
> >> Elliotte Rusty Harold
> >> [email protected]
> >>
> >> ---------------------------------------------------------------------
> >> To unsubscribe, e-mail: [email protected]
> >> For additional commands, e-mail: [email protected]
> >>
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [email protected]
> > For additional commands, e-mail: [email protected]
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to