9cd473d7247e67541b208d7a49beae5f4c3d6d99: hgmo: serve AWS clients bundles from local S3 region (bug 1249197); r=fubar, dminor
Gregory Szorc <gps@mozilla.com> - Fri, 19 Feb 2016 12:42:38 -0800 - rev 362299
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hgmo: serve AWS clients bundles from local S3 region (bug 1249197); r=fubar, dminor This commit modifies the hgmo extension to inspect the source IP address of HTTP requests for bundle clone and clone bundles manifests and compare against known AWS IP ranges. If the request is coming from an AWS region that we have S3 URLs for (identified by the presence of "ec2region=" in the manifest metadata), we filter the manifest to only advertise URLs in the same region. This guarantees an intra-region transfer without any client configuration. This is fast and free since intra-region S3 transfers don't cost anything! In addition, we also reorder the manifest to advertise the stream clone bundles first. We recommend clients in automation apply stream clone bundles (over the default gzip bundles) because they are the fastest mechanism to clone repositories (a stream clone bundle can apply at >80 MB/s). Tests for the new features have been added to the hgserver tests because we have existing tests that run the code for generating manifests and we want to use realistic values for testing. Documentation for the bundle hosting has been updated to reflect the change in behavior. MozReview-Commit-ID: 2ZaoJpaiqBU
d4f71a509bacf13ca4ccdcf03ba1a161cb45200b: ansible/hg-web: install ipaddress Python package (bug 1249197); r=fubar
Gregory Szorc <gps@mozilla.com> - Fri, 19 Feb 2016 12:41:14 -0800 - rev 362298
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
ansible/hg-web: install ipaddress Python package (bug 1249197); r=fubar Python 2 doesn't have a nice API for parsing IP addresses and doing netmask foo. Python 3.3 introduced the "ipaddress" package. There is a backport for Python 2, which we install in this commit. MozReview-Commit-ID: 85JSXoraBlO
9741cca32ff47892f7ca5e71e680fa586d1554d7: ansible/hg-web: install AWS IP ranges JSON file (bug 1249197); r=fubar
Gregory Szorc <gps@mozilla.com> - Fri, 19 Feb 2016 12:41:08 -0800 - rev 362297
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
ansible/hg-web: install AWS IP ranges JSON file (bug 1249197); r=fubar We'll need this to examine source IPs to see if they are from AWS. File obtained from https://ip-ranges.amazonaws.com/ip-ranges.json and checked in unmodified. MozReview-Commit-ID: Emq8DqXrUkY
d6d96bde19fd8c353e3ba61301c7b304d70deb13: mozreview: make double-clicking to open a comment optional per user (bug 1246769); r=mdoglio
Mike Conley <mconley@mozilla.com> - Thu, 18 Feb 2016 10:04:10 -0500 - rev 362296
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: make double-clicking to open a comment optional per user (bug 1246769); r=mdoglio MozReview-Commit-ID: 9cZ83ZehALQ
fbbac518215ad0a639a52f1cb9e34ce3413b7aa5: mozreview: make autoland selenium tests check autoland button title (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Thu, 11 Feb 2016 15:00:42 -0500 - rev 362295
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: make autoland selenium tests check autoland button title (bug 1243530) r=mdoglio Right now we check the opacity of the autoland button to see if it is enabled. If we check the title instead we'll get better errors when something goes wrong. MozReview-Commit-ID: 3GMw5a7SfRl
c6beb77d5a14994eeb1814283cd9e27ba39b0733: mozreview: fix title for autoland button on discarded review (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Thu, 18 Feb 2016 08:40:13 -0500 - rev 362294
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: fix title for autoland button on discarded review (bug 1243530) r=mdoglio MozReview-Commit-ID: D9ZXDOPncnP
91efb7ecba60e4cad563fdd69d294819befa3945: mozreview: use new hostingservice fields to disable autoland in UI (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Thu, 11 Feb 2016 06:47:54 -0500 - rev 362293
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: use new hostingservice fields to disable autoland in UI (bug 1243530) r=mdoglio MozReview-Commit-ID: LKaVIRB4wAf
f9b4d74d9a5dc27a16ae9f512adf7a332ca81293: testing: fix missing import in unittest.py (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Thu, 18 Feb 2016 08:42:45 -0500 - rev 362292
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
testing: fix missing import in unittest.py (bug 1243530) r=mdoglio MozReview-Commit-ID: DiSWl1N8fQw
e4808063d4521fa938f085eb5ee0ecaee966a11e: mozreview: reject autoland requests for disabled repositories (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Thu, 11 Feb 2016 15:29:17 -0500 - rev 362291
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: reject autoland requests for disabled repositories (bug 1243530) r=mdoglio MozReview-Commit-ID: 1zTsh1lM2Rx
c3e58764d22c342397cb70a4609a15a1a3490773: mozreview: add fields to disable autoland in custom hosting service (bug 1243530) r=mdoglio
Dan Minor <dminor@mozilla.com> - Tue, 02 Feb 2016 15:28:00 -0500 - rev 362290
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: add fields to disable autoland in custom hosting service (bug 1243530) r=mdoglio MozReview-Commit-ID: EtvIW0rsPPZ
8f4dfd0518b2575239bebbbe32cfc090a4febd45: hgmo: prune some keys from the returned JSON (bug 1248850); r=dminor
Gregory Szorc <gps@mozilla.com> - Wed, 17 Feb 2016 10:17:28 -0800 - rev 362289
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hgmo: prune some keys from the returned JSON (bug 1248850); r=dminor This is a very hacky way to avoid a templating issue due to changelogtag missing. Fixing the underlying issue requires handling special return values from Mercurial's internal templating API and I have little desire to do that because those APIs are fragile. It is easier to just remove the output that calls into them to avoid the problem. MozReview-Commit-ID: BmcB5LnfJkj
09148d3c7d5a3ace7dacdd2de4cb1c1f9678128f: hgmo: web command for exposing automation relevance (bug 1247802); r=dminor
Gregory Szorc <gps@mozilla.com> - Tue, 16 Feb 2016 14:43:24 -0800 - rev 362288
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hgmo: web command for exposing automation relevance (bug 1247802); r=dminor I had planned to expose JSON info for automation relevance via Mercurial's built-in JSON API. Unfortunately, the search functionality of that API is not yet implemented. Bug 1247322 is on file to track that. I looked into implementing it, but it would be a few times more work than implementing a trivial one-off web command that exposes the same data. This commit introduces a web command for exposing the results of an automationrelevant() revset query for a revision. It calls into the hgweb functions for populating changelist entries, which means our extensions are called to populate custom metadata. So we get pushlog info in the output. Yay. MozReview-Commit-ID: BzXUyAoCJQo
407a08387e0936a2415be1ed5585c68154085b2b: hgmo: add a revset to return changesets relevant to automation (bug 1247802); r=dminor
Gregory Szorc <gps@mozilla.com> - Tue, 16 Feb 2016 14:42:59 -0800 - rev 362287
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
hgmo: add a revset to return changesets relevant to automation (bug 1247802); r=dminor Today, Firefox's automation isn't very intelligent about what jobs it schedules for each push. It is overly aggressive and tends to schedule everything all the time. This doesn't scale. In order to schedule only what actually needs to be scheduled, you need to know "what changed since last time." Build systems answer this question by consulting the filesystem: if an mtime is newer than its output file, the output file is stale and needs to be rebuilt. We can't rely on this in automation because there are several build machines and build machines tend to be frequently wiped. This means the state of the mtimes on the filesystem can't be trusted and will yield incorrect results for determining what changed. Fortunately we have a versioned filesystem in the form of version control. With version control, we can compute the set of files that changed between 2 revisions very easily. The trick is identifying the 2 revisions. Firefox automation schedules jobs from pushes to version control. When a new push comes in, the revision at the head of that push is checked out and built, etc. The head of the push is 1 revision. But we still need an ancestor revision to compare it to so we can get the set of files that changed. We can't simply select the ancestor of the first revision in the push because of the monolithic, many-headed Try repository. If you push 4 commits to Try then rewrite the tip-most commit and push to Try again, that second push will only contain a single commit and selecting the ancestor of that commit will only return 2 of the 4 commits relevant to that Try push! A solution to this problem is to consult Mercurial's phases information to supplement the changesets in a push with their non public ancestors. This commit introduces the "automationrelevant" revset. Given a revision, it will select all revisions relevant to that revision for purposes of running build automation. The revset is currently equivalent to: nodesfrompush(REV) | (::REV & not public()). Although we don't have a "nodesfrompush" or equivalent revset. We /could/ implement such a revset. However, even if we had one, I'd rather automation not construct their own revset because it is a bit more complicated and the behavior of this revset may change over time. For example, I anticipate there will eventually be an inbound-like non-publishing repo. For this repo, "automationrelevant" should *not* extend with draft changesets. In the future, we can teach the revset to behave differently when executed on this type of repo. After this lands, the plan is to teach automation to call a JSON API that runs this revset and returns changeset metadata, from which it can calculate the set of changed files and make intelligent scheduling decisions. MozReview-Commit-ID: BYYdV2m70cB
39a6434928cda7bc27db261a3f567b6c6e61ac79: Bug 1247838 - Fix running git mozreview from a subdirectory. r=gps
Mike Hommey <mh@glandium.org> - Fri, 12 Feb 2016 14:40:03 +0900 - rev 362286
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
Bug 1247838 - Fix running git mozreview from a subdirectory. r=gps Bug 1246998 broke it, and the push code in cinnabar assumes it's running from the toplevel directory, which is a fine assumption in the git-remote-hg helper, but not in the git cinnabar command. Until cinnabar itself is fixed, work around it. While here, add a test case for worktrees.
79f7e3cb540975eeb05ddd4a7d7db083cf0d8f27: mozreview: discarded child reviews shouldn't break bmo integration (bug 1243021) r=mdoglio
byron jones <glob@mozilla.com> - Tue, 16 Feb 2016 01:02:36 +0800 - rev 362285
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozreview: discarded child reviews shouldn't break bmo integration (bug 1243021) r=mdoglio Fixes a crash triggered by a child review request being marked as discarded. MozReview-Commit-ID: 5AvXwTAmXkE
e308dcce64dc4162b89e2d63c943a41b95135ff2: Bug 1247950 - Fix error when no pushlog data are available; r=gps
Calixte Denizet <cdenizet@mozilla.com> - Fri, 12 Feb 2016 12:35:38 -0800 - rev 362284
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
Bug 1247950 - Fix error when no pushlog data are available; r=gps
02aec36de661d6dd4e687e9e14fcd8663e76638e: autoland: set a max height on the commit confirmation dialog (bug 1247569) r=dminor
byron jones <glob@mozilla.com> - Fri, 12 Feb 2016 00:46:58 +0800 - rev 362283
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
autoland: set a max height on the commit confirmation dialog (bug 1247569) r=dminor djblet's modal box doesn't restrict the height of content to the window, resulting in a dialog with unreadable content when there are a lot of commits. Unfortunately fixing this correctly by bounding the dialog to the window isn't possible with RB css, so we just cap the maximum height at 600px. MozReview-Commit-ID: HlTCNztV8ci
58227cb3472721e6c361a692688b5c47d8b0441a: Bug 1247172 - Do not use an actual remote for git mozreview r=gps
Mike Hommey <mh@glandium.org> - Thu, 11 Feb 2016 09:01:37 +0900 - rev 362282
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
Bug 1247172 - Do not use an actual remote for git mozreview r=gps
a1ec4b577c092a9757d27f2d202f687de3ca262c: Bug 1246998 - Take worktrees into account in git mozreview r=gps
Mike Hommey <mh@glandium.org> - Thu, 11 Feb 2016 12:51:03 +0900 - rev 362281
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
Bug 1246998 - Take worktrees into account in git mozreview r=gps
69f6e5839c4c905ad5a419ddd7660806b0caebf6: mozext: don't attempt to pull pushlog when pulling via ssh:// (bug 1242281); r=smacleod
Gregory Szorc <gps@mozilla.com> - Wed, 10 Feb 2016 13:49:50 -0800 - rev 362280
Push 16998 by rwood@mozilla.com at Mon, 02 May 2016 19:42:03 +0000
mozext: don't attempt to pull pushlog when pulling via ssh:// (bug 1242281); r=smacleod Read the inline comment for why we do this. MozReview-Commit-ID: Fddeqmu8UUB
(0) -300000 -100000 -30000 -10000 -3000 -1000 -300 -100 -50 -20 +20 +50 +100 +300 +1000 +3000 +10000 +30000 +100000 +300000 tip