Naively (i.e., without knowing anything about the implications :-) I think I agree that options #2 sounds like the way to go.
regards, Karl On Tue, Nov 21, 2017 at 10:45 AM, Konrad Windszus <konra...@gmx.de> wrote: > Definitely option 2. > Konrad > >> On 21. Nov 2017, at 10:27, Radu Cotescu <r...@apache.org> wrote: >> >> Hi, >> >> It looks like the org.apache.sling.starter.startup bundle breaks the way >> some integration tests check that an instance is ready, since the Servlet >> Filter registered by this bundle returns a 200 for any accessed path. >> >> The mentioned tests look for a path in the repo by doing a GET request and >> expecting a 200 response code (this functionality is also provided by >> org.apache.sling.testing.junit.rules.SlingInstanceRule#getAdminClient()#waitUntilExists()). >> >> There are obviously at least two solutions to this problem: >> >> 1. tests shouldn't rely on a 200 response code from various paths until >> they first check that accessing index.html returns the proper page, by >> looking for the <!-- Do not remove this comment, used for Launchpad >> integration tests --> comment; OR >> >> 2. the Servlet Filter registered by org.apache.sling.starter.startup should >> in fact return a 503 - Service Unavailable, with the current response body >> mentioning that Apache Sling is starting >> >> I'd definitely prefer the 2nd option, since IMO we should not include >> testing-related content in the default Apache Sling Starter Application and >> 503 is an appropriate answer for a starting instance. >> >> Thoughts? >> >> Thanks, >> Radu >> >> On Wed, 4 Oct 2017 at 12:03 Bertrand Delacretaz <bdelacre...@apache.org> >> wrote: >> >>> On Wed, Oct 4, 2017 at 12:00 PM, Carsten Ziegeler <cziege...@apache.org> >>> wrote: >>>> If we can agree on starter, I guess we should change the artifact id >>> >>> +1 >>> >>> -Bertrand >>> > -- Karl Pauls karlpa...@gmail.com
