Hi Ben, Thanks!
I will add the test classe and include this PR the release 1.5.0. regards, François [email protected] Le 14/01/2020 à 07:48, Benjamin Marwell a écrit : > Thank YOU for looking into this! > > In case you want to adapt this as an integration test case, there are other > ways to start openLiberty. > > One is through Arquillian, another by using test containers. You can find > examples of both methods here: > https://aguibert.github.io/openliberty-cheat-sheet/#_integration_testing > > > > On Mon, 13 Jan 2020, 22:40 Brian Demers, <[email protected]> wrote: > >> Thanks Ben! this is very helpful! >> >> On Mon, Jan 13, 2020 at 3:06 PM Benjamin Marwell <[email protected]> >> wrote: >> >>> Hello Brian, >>> >>> I just figured out that I opened two PMRs at IBM. One just recently, >>> the other one about a year and a half ago. That one was closed because >>> "the NPE is just a wrong error message" and therefore was converted to >>> a RFE. >>> This way is almost impossible to figure out what went wrong. I think >>> the underlying problem is in fact Apache CXF. >>> >>> Here's a working example: >>> https://github.com/bmhm/shiro-openliberty-npe >>> >>> just run `mvn liberty:dev` and press Enter to execute the IT. >>> >>> You will be able to see the stacktrace in STDOUT and it is also >>> available in the README file. >>> >>> Fix: >>> 1. Remove shiro-servlet-plugin >>> 2. Insert the missing parts into a new file web.xml and add the two lines >>> of PR. >>> >>> HTH >>> >>> You can use the code as an IT for apache shiro if you like. >>> >>> >>> Am Mo., 13. Jan. 2020 um 19:14 Uhr schrieb Brian Demers >>> <[email protected]>: >>>> Thanks Ben! >>>> >>>> Can you include the NPE stacktrace too? That might help us write a test >>>> (and maybe push a fix into OpenLiberty, even if it's just a better >>>> exception) >>>> >>>> On Sat, Jan 11, 2020 at 10:15 AM Benjamin Marwell <[email protected]> >>>> wrote: >>>> >>>>> Hi, >>>>> >>>>> I would like to see my PR [1] merged, because this causes a lit of >>> trouble >>>>> for me and possibly other users. I do not think there are any side >>> effects, >>>>> but another review might still be beneficial. >>>>> >>>>> Also, one could write an integration test like this: start an >> affected >>>>> application server, e.g. OpenLiberty and run a htto request targeted >>> at a >>>>> servlet whose method uses such a signature: >>>>> >>>>> public void getAction(final @Suspended asyncResponse >>> asyncResponse); >>>>> Any help would be appreciated, although it could probably merged >> right >>> away >>>>> even without such a test case. >>>>> >>>>> Thanks, >>>>> Ben >>>>> >>>>> [1] https://issues.apache.org/jira/browse/SHIRO-735 >>>>> <https://issues.apache.org/jira/browse/SHIRO-735> >>>>>
