4ca0f73b40c352df7e149daaaa38242b8697c297: qimportbz: exit gracefully if there are no patches to import (bug 561306)
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 13:52:39 +0000 - rev 360605
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: exit gracefully if there are no patches to import (bug 561306) Previously, if: * the bug was inaccessible * there were no patches attached * the only attached patches were obsolete and the user declined to use them ...we'd end up with an unhandled TypeError exception due to mq trying to use .read() on our None return value. With this change, we exit more gracefully, with an "abort: foo" instead.
5c3aa12de69b22602ed39b9686a8cfd1740fda34: qimportbz: cleanup the "rename patch" prompt message (bug 980986)
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 12:26:22 +0000 - rev 360604
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: cleanup the "rename patch" prompt message (bug 980986) This slightly shortens the string and displays it to the user over two lines, for improved readability. We also no longer use the slightly misleading name of 'prompt' for the message, since it's not actually showing the prompt yet.
8043dea8a1af40a4f8c342446c01bffa7586313c: qimportbz: don't rename patches to 'y' if no filename entered (bug 980986)
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 12:16:46 +0000 - rev 360603
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: don't rename patches to 'y' if no filename entered (bug 980986) Currently if return is pressed prior to entering a patch name, the patch gets renamed to 'y'. With this change the existing name is kept, which causes us to go around the while loop again & have another chance to enter a name.
5d7f5b1eb7271108a7f7f1a0f490fbd8a890de09: qimportbz: avoid filename collisions with the temporary patch file (bug 980986)
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 12:03:00 +0000 - rev 360602
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: avoid filename collisions with the temporary patch file (bug 980986) Prior to this patch, if an imported patch had a filename that matched the bug number (and the user hadn't overriden patch_format in their local config), an error would be shown saying that the patch already existed with the same name, even when the file did not exist prior to running the import. This is due to the temporary file (created only for the first patch in the sequence, later patches don't have one) having the same name as our target filename. To solve this, we add a suffix to the target filename, which will allow both this and subsequent imports to succeed without conflicts with the temporary file.
98eeed7a40ad1107057a725112f838dae6cd9ea5: qimportbz: pyflakes, pep8, comment & whitespace cleanup
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 10:04:17 +0000 - rev 360601
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: pyflakes, pep8, comment & whitespace cleanup
39a2e8248e8ac5be7c1f4aac0d506c65ed782354: qimportbz: Remove support for Python older than v2.6
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 10:00:37 +0000 - rev 360600
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: Remove support for Python older than v2.6
1212e433a461e890b315da533102cf68fa17bb78: qimportbz: Remove support for Mercurial older than v1.9
Ed Morley <emorley@mozilla.com> - Fri, 13 Mar 2015 09:54:44 +0000 - rev 360599
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
qimportbz: Remove support for Mercurial older than v1.9
3f82a2eceeea1552a07e0d5f6106478f1299653b: docker: support building with nocache
Gregory Szorc <gps@mozilla.com> - Tue, 17 Mar 2015 10:44:05 -0700 - rev 360598
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
docker: support building with nocache
f60e275cc9af4e6ac5d7be0183db662f3a5c1d71: docker: allow %include /
Gregory Szorc <gps@mozilla.com> - Tue, 17 Mar 2015 10:40:35 -0700 - rev 360597
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
docker: allow %include / This wasn't working before because we were using prefix matching.
83f107c3a8c82d90cd756ca73cdb1739e93fe512: hgmo: factor ldap code into generic and reusable module
Gregory Szorc <gps@mozilla.com> - Mon, 16 Mar 2015 15:16:55 -0700 - rev 360596
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hgmo: factor ldap code into generic and reusable module The LDAP code was originally written as part of the hgmo code. Since we will soon use the LDAP server as part of the MozReview environment, this code needs to be usable by multiple consumers. So, we extract the code into its own generic module.
ff64f5d5fbb3dc7537b58a5cf9b877fcd45310e0: reviewboard: always push via ssh://
Gregory Szorc <gps@mozilla.com> - Mon, 16 Mar 2015 13:07:53 -0700 - rev 360595
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
reviewboard: always push via ssh:// We will be moving the Mercurial server in the test environment to a Docker container so the test environment behaves more like production. Prepare for that by converting tests to push via ssh:// instead of http://.
af75edd642f7e6ec8a7b40e810a252b7b924601a: docker: remove references to scan_includes
Gregory Szorc <gps@mozilla.com> - Mon, 16 Mar 2015 17:07:35 -0700 - rev 360594
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
docker: remove references to scan_includes This argument was removed. Remove use of it.
65ea7e79670d4351e5fde87241c6bad4095865c1: testing: start an LDAP server as part of MozReview
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 18:25:23 -0700 - rev 360593
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: start an LDAP server as part of MozReview We want the MozReview test environment to look as similar to production as possible. To do that, we need to run a Mercurial server that's similar to production. To do that, we need an LDAP server to authenticate SSH requests against. This patch starts a Mozilla flavored LDAP server as part of the MozReview cluster environment. We currently don't do anything with the server. But this is a pretty large patch already, so stopping here makes sense.
ee099c5c42ad382a917cabbbe3f3bd24b2468831: hghooks: remove old push_repo.py hook
'Ben Kero' - Mon, 16 Mar 2015 21:36:42 +0000 - rev 360592
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hghooks: remove old push_repo.py hook Bug 1143869 This hook doesn't seem to be used anymore, as hinted at by the files it references not existing. Proposed deleting.
d0f2337f16fd061cb672ccfd32d2ca8261bc4a15: testing: build all Docker images during create-test-environment
Gregory Szorc <gps@mozilla.com> - Mon, 16 Mar 2015 13:04:20 -0700 - rev 360591
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: build all Docker images during create-test-environment Before, we were only building images used by MozReview. Change that so we don't build images silently the first time we go to run tests.
4f789aa86966c766bd28d5e1a97b9aaaab902752: testing: allocate more ports to the Mercurial test runner
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 17:39:18 -0700 - rev 360590
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: allocate more ports to the Mercurial test runner
352fcef6c2593896184b17a363e4b7886800e162: testing: return a dict from build_mozreview
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 16:16:49 -0700 - rev 360589
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: return a dict from build_mozreview A tuple made sense when we had 2 images. We now have 5 and are planning more. Convert to a dict to preserve our sanity.
6f39af8570a73ac9b5b6d3f186c2f232d1505b41: testing: build all images concurrently
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 15:52:00 -0700 - rev 360588
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: build all images concurrently Previously, build_all_images executed the build_mozreview and build_hgmo functions concurrently to do building. This worked as long as there was no overlap between images between the two environments. Upcoming work to integrate a Dockerized Mercurial server into the MozReview environment meant that the ldap image was being built by both environments. This caused race conditions and double building. We solve this problem by refactoring the build_* APIs to accept a dict of already-built images. The API for building images now also only builds images that aren't known. build_all_images() is now an explicit list of images to build. This enables us to start using the ldap image in both environments.
372a2a643c750d80362ef17f7a5ee24f5b929efc: testing: remove add_vct and scan_includes from image building
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 15:36:22 -0700 - rev 360587
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: remove add_vct and scan_includes from image building scan_includes mode is now the default. Consistency is important for understanding. It also makes code simpler since we don't need to know what flavor a particular image is when building it.
77679c09f2d3f284f4d3ea46ca3fa9c2e1848d0a: autoland: convert Docker images to use scan_includes
Gregory Szorc <gps@mozilla.com> - Fri, 13 Mar 2015 15:17:13 -0700 - rev 360586
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
autoland: convert Docker images to use scan_includes add_vct is now not utilized anywhere meaningful. We only add the requested files to the builder context.
(0) -300000 -100000 -30000 -10000 -3000 -1000 -300 -100 -50 -20 +20 +50 +100 +300 +1000 +3000 +10000 +30000 +100000 +300000 tip