Based on the feedback in this week's meeting, I intend to go ahead and move this straight to a vote thread.
On Wed, Jul 2, 2025 at 10:23 AM James Fredley <jamesfred...@apache.org> wrote: > I believe this is the quickest and likely best way to ensure GSP in Grails > 7 works the way it did in Grails 5 and 6. > > It addresses the long term open issues with Sitemesh 3 updates for Grails > 7. We can revisit Sitemesh 3 for Grails 8. > > It also fixes failing tests that we didn't know were caused by Sitemesh 3 > changes made before the majority of dependencies were updated for Grails 7 > in Summer and Fall 2024. > > James Fredley > > On 2025/07/02 11:38:52 James Daugherty wrote: > > Hi Everyone, > > > > I started investigating a ticket related to forwarding not working in a > > Grails controller on 7.0.0-M4. As part of that I made this discovery: > > > https://github.com/apache/grails-core/issues/14193#issuecomment-3017157073 > > It seems the DispatcherServlet from Spring is designed to handle > multiple > > calls methods that cause the response to render and close. The > dispatcher > > chooses the right one by the committed state of the response. > > > > From my investigation, the Sitemesh 3 implementation appears to use a > > filter to render its output after the DispatcherServlet is processed - so > > the response state is never committed. I think we could update the > > GrailsDispatcherServlet to handle the various scenarios that the > > DispatcherServlet handles, but I'm not sure this is practical for Grails > 7 > > given the delays we have already had. We would also be deviating from > how > > Spring MVC controllers are rendered and this logic doesn't appear that > > simple. Moreover, we still have several outstanding tickets related to > the > > sitemesh 3 upgrade: > > > > * Apr 2025 - https://github.com/apache/grails-core/issues/14193 - > > forwarding from a controller > > * Jan 2025 - https://github.com/apache/grails-core/issues/13988 - nested > > layouts > > * Feb 2005 - https://github.com/apache/grails-core/issues/14197 - $raw > in > > titles > > * Nov 2024 - https://github.com/apache/grails-core/issues/13849 - > layout by > > convention > > > > We have reached out to the developer of Sitemesh 3 on most of these > tickets > > (via ticket comments and direct slack messages over the past few months); > > we have yet to receive any work or progress on them. The oldest of which > is > > 8+ months old. I'm skeptical these can be finished in a reasonable time > > frame. > > > > > > Given the lack of support, the multiple issues found by users, and the > > current state of Grails 7 being delayed by these tickets, I would like to > > propose we revert back to Sitemesh 2.6.0. Assuming we confirm all of > these > > issues are fixed, we would only have the remaining license, asset plugin > > release, & ASF release related issues to fix before we could release > Grails > > 7. > > > > > > I have gone ahead and reverted back to Sitemesh 2.6.0 in the issue14193 > > branch (https://github.com/apache/grails-core/tree/issue14193). As > part of > > switching back to v2, I discovered the following is now fixed: > > * async rendering > > * async error handling > > * layout specified by a controller property > > * layout specified in controller property applied to a GSP that does not > > contain a root html tag > > > > It seems the sitemesh3 upgrade was performed during a time when the > project > > wasn't running all of its tests, so we are only now learning the scope of > > its impact. I think we can still be open to switching to Sitemesh 3 in > > Grails 8. Especially if we get the necessary support, but otherwise, we > > can continue with 2.x. > > > > Please note as part of this branch revert, I've also taken a closer look > at > > the Sitemesh licensing requirements and made a first pass being > compatible > > with the Sitemesh license. We'll still have to follow-up with ASF legal > to > > ensure we're compliant since the license is a derivative of the ASF > > license. > > > > -James > > >
