Info

See dev-only comment with patch build links, but as far as I understand it, changes to SConstruct should disable compile bypass. See also EVG-12714.

Top User Comments

xgen-internal-githook commented on Wed, 12 Aug 2020 17:10:40 +0000: Author: {'name': 'David Bradford', 'email': 'david.bradford@mongodb.com', 'username': 'dbradf'} Message: SERVER-50078: Disable bypass_compile (cherry picked from commit b1d88bac122fcadedcb0be3798872770de9c732a) (cherry picked from commit 4a23cd035cc14fe281f265e7d42b032f9ca8434f) (cherry picked from commit 1f3af36a3663a6b85fec18a14dab756a54574243) Branch: v4.0 https://github.com/mongodb/mongo/commit/b664d3bf5ab7642b9a0ef3289e666ed1043572e9 xgen-internal-githook commented on Wed, 12 Aug 2020 15:38:27 +0000: Author: {'name': 'David Bradford', 'email': 'david.bradford@mongodb.com', 'username': 'dbradf'} Message: SERVER-50078: Disable bypass_compile (cherry picked from commit b1d88bac122fcadedcb0be3798872770de9c732a) (cherry picked from commit 4a23cd035cc14fe281f265e7d42b032f9ca8434f) Branch: v4.2 https://github.com/mongodb/mongo/commit/1f3af36a3663a6b85fec18a14dab756a54574243 xgen-internal-githook commented on Wed, 12 Aug 2020 15:13:38 +0000: Author: {'name': 'David Bradford', 'email': 'david.bradford@mongodb.com', 'username': 'dbradf'} Message: SERVER-50078: Disable bypass_compile (cherry picked from commit b1d88bac122fcadedcb0be3798872770de9c732a) Branch: v4.4 https://github.com/mongodb/mongo/commit/4a23cd035cc14fe281f265e7d42b032f9ca8434f xgen-internal-githook commented on Wed, 12 Aug 2020 14:45:10 +0000: Author: {'name': 'David Bradford', 'email': 'david.bradford@mongodb.com', 'username': 'dbradf'} Message: SERVER-50078: Disable bypass_compile Branch: master https://github.com/mongodb/mongo/commit/b1d88bac122fcadedcb0be3798872770de9c732a david.bradford commented on Wed, 12 Aug 2020 14:14:00 +0000: I see what is happening. The changes are being applied to the patch build differently than they used to be. So when bypass_compile looks at what files have changed, it is getting an empty list which makes it think it is safe to bypass_compile. For the short-term, I am going to just disable bypass_compile. Longer-term, I think having the user specify they want bypass_compile enabled when submitting their patch build makes the most sense. Having bypass_compile for things like test-only changes can save a lot of time, but bugs in the scripts are causing unacceptable mistakes. For this, we would likely need something like parameterized builds in Evergreen. brooke.miller commented on Fri, 7 Aug 2020 14:39:03 +0000: If this becomes more complex, we should consider ticketing this work out further.

Steps to Reproduce