I made it a JUnit assume so it would ignore the test when the class
isn't found. I think I made a Hamcrest matcher for it.

On Wed, 26 Jun 2019 at 23:04, Ralph Goers <[email protected]> wrote:
>
> I tried porting the changes you made back to release-2.x and it compiles, but 
> it still fails in the unit tests as the reflection fails since the class 
> isn’t present. How did you solve that?
>
> Ralph
>
> > On Jun 26, 2019, at 1:12 PM, Ralph Goers <[email protected]> wrote:
> >
> > Nevermind. I see you used reflection. I am not sure why I hadn’t thought of 
> > doing that.
> >
> > Ralph
> >
> >> On Jun 26, 2019, at 1:08 PM, Ralph Goers <[email protected]> 
> >> wrote:
> >>
> >> How did you get around the compile problem in master?
> >>
> >> Ralph
> >>
> >>> On Jun 26, 2019, at 12:36 PM, Matt Sicker <[email protected]> wrote:
> >>>
> >>> I fixed that sun.reflect compile error in master, but didn't backport
> >>> it to release-2.x. There are still some other compile errors besides
> >>> if you compile with Java 11.
> >>>
> >>> As for why you get different results when running rat with or without
> >>> the rat profile activated, it's because our rat config is different in
> >>> the rat profile compared to the default one. I think our config is
> >>> wrong there, and I typically rediscover that any time I do a release.
> >>>
> >>> On Wed, 26 Jun 2019 at 13:15, Ralph Goers <[email protected]> 
> >>> wrote:
> >>>>
> >>>> I have wondered that for a long time. In fact, if you run “mvn compile” 
> >>>> followed by “mvn test” you will see
> >>>>
> >>>> [INFO] Changes detected - recompiling the module!
> >>>>
> >>>> So obviously something is there that tries to detect changes. It just 
> >>>> doesn’t seem to work.
> >>>>
> >>>> Ralph
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>> On Jun 26, 2019, at 9:32 AM, Gary Gregory <[email protected]> 
> >>>>> wrote:
> >>>>>
> >>>>> On Wed, Jun 26, 2019 at 12:24 PM Ralph Goers 
> >>>>> <[email protected]>
> >>>>> wrote:
> >>>>>
> >>>>>> Those are mine as well.
> >>>>>>
> >>>>>> Unfortunately, if you use the toolchains plugin to specify the compiler
> >>>>>> version then all the tests are also going to run with that Java 
> >>>>>> version.
> >>>>>> Since there is no getting around having sun.reflect.Reflection be used
> >>>>>> prior to Java 9 there is no way to use Java 11 to compile and test 
> >>>>>> Log4j
> >>>>>> API.  That only leaves you with the option of creating a phony jar. But
> >>>>>> that will have problems too as it will probably be used by the unit 
> >>>>>> tests.
> >>>>>>
> >>>>>> The only other option is to create a separate module for the unit 
> >>>>>> tests.
> >>>>>>
> >>>>>
> >>>>> But why does Maven recompile when nothing has changed? That's the real
> >>>>> problem, I ran a first build on Java 8, then all I want to do is run the
> >>>>> tests on Java 11.
> >>>>>
> >>>>> Gary
> >>>>>
> >>>>>>
> >>>>>> Ralph
> >>>>>>
> >>>>>>> On Jun 26, 2019, at 9:00 AM, Gary Gregory <[email protected]>
> >>>>>> wrote:
> >>>>>>>
> >>>>>>> On Wed, Jun 26, 2019 at 11:02 AM Ralph Goers 
> >>>>>>> <[email protected]
> >>>>>>>
> >>>>>>> wrote:
> >>>>>>>
> >>>>>>>> I don’t run the build with Java 11, although I do have projects that 
> >>>>>>>> use
> >>>>>>>> it that are on Java 11. It seems that log4j-api is failing when
> >>>>>> compiling
> >>>>>>>> with Java 11 because sun.reflect.Reflection was removed in Java 9.  I
> >>>>>> guess
> >>>>>>>> we need to modify the toolchains configuration to tell the main 
> >>>>>>>> modules
> >>>>>> to
> >>>>>>>> use only Java 8 to compile.  The only other option would be to 
> >>>>>>>> create a
> >>>>>> jar
> >>>>>>>> that has that class in it and specify it as a provided dependency.
> >>>>>>>>
> >>>>>>>
> >>>>>>> My goal is to be able to answer the question: "Do all the tests pass 
> >>>>>>> on
> >>>>>>> Java 11?"
> >>>>>>>
> >>>>>>> I am OK for this release with not being able to _build_ on Java 11 
> >>>>>>> but I
> >>>>>>> certainly should be able to run all tests with a plain old 'mvn test'.
> >>>>>>>
> >>>>>>> FWIW, at work, our platform requirements are Java 8 and 11.
> >>>>>>>
> >>>>>>> Gary
> >>>>>>
> >>>>>>
> >>>>
> >>>>
> >>>
> >>>
> >>> --
> >>> Matt Sicker <[email protected]>
> >>>
> >>
> >>
> >>
> >
> >
> >
>
>


-- 
Matt Sicker <[email protected]>

Reply via email to