From mark.karpov at tweag.io Wed Apr 11 08:40:27 2018 From: mark.karpov at tweag.io (Karpov, Mark) Date: Wed, 11 Apr 2018 15:40:27 +0700 Subject: [GHC DevOps Group] Stackage HEAD, now working Message-ID: Hello, Last year there was an initiative to execute Stackage nightly build plans with development versions of GHC and by comparing results that we obtain with different commits we could detect how they affect packages in our ecosystem. Thus we potentially can detect regressions and other suspicious changes. * The original blog post introducing the initiative: https://www.tweag.io/posts/2017-10-27-stackage-head.html * A talk from HIW'17: https://www.youtube.com/watch?v=UAI-mplXUkY Recently we resumed working on this and actually I'm happy to announce that Stackage HEAD is functional now. In fact, a bit unexpectedly, it detected something today: https://ghc.haskell.org/trac/ghc/ticket/15022 This is not necessarily a bug or regression, but still fewer test suites of stm-delay build at 3cfb12d8adac37e5565d66fd173e4648cc041e65 than before at 5161609117c16cb7b29b2b8b1cd41e74341d4137 which is a suspicious change to Stackage HEAD build report manager. There is a short description how it works here: https://github.com/tweag/stackage-head#how-it-works The logic that determines if a change in status of a package is suspicious is here: https://github.com/tweag/stackage-head/blob/master/Stackage/HEAD/BuildDiff.hs#L61-L94 And the entire source code of the tool is available here: https://github.com/tweag/stackage-head As you can see, a package can either fail to build, or be unreachable (due to failing dependencies) or it can succeed in which case we record number of succeeding/failing test suites. Currently statistics look like this (with nightly-2018-04-05): * Failing packages: 5 * Unreachable packages: 794 * Packages that build: 897 The build takes about one hour. Circle CI logs are available here: https://circleci.com/gh/tweag/stackage-head I configured this to run 4 times per day for a start. I'll receive email notifications on failures which I plan to convert to trac tickets manually (for now.). I could not find a way to make Circle CI send emails to other people like Ben, but I see there are various hook available, so we could use some of them if desirable. Best, Mark -------------- next part -------------- An HTML attachment was scrubbed... URL: From m at tweag.io Wed Apr 11 08:45:46 2018 From: m at tweag.io (Boespflug, Mathieu) Date: Wed, 11 Apr 2018 10:45:46 +0200 Subject: [GHC DevOps Group] Stackage HEAD, now working In-Reply-To: References: Message-ID: Hi folks, Stackage HEAD is currently administered in the tweag GitHub org, due to practical constraints. But I'd encourage moving it over to the ghc or haskell orgs if the community finds this useful. Best, -- Mathieu Boespflug Founder at http://tweag.io. On 11 April 2018 at 10:40, Karpov, Mark wrote: > Hello, > > Last year there was an initiative to execute Stackage nightly build plans > with development versions of GHC and by comparing results that we obtain > with different commits we could detect how they affect packages in our > ecosystem. Thus we potentially can detect regressions and other suspicious > changes. > > * The original blog post introducing the initiative: > https://www.tweag.io/posts/2017-10-27-stackage-head.html > * A talk from HIW'17: https://www.youtube.com/watch?v=UAI-mplXUkY > > Recently we resumed working on this and actually I'm happy to announce > that Stackage HEAD is functional now. In fact, a bit unexpectedly, it > detected something today: > > https://ghc.haskell.org/trac/ghc/ticket/15022 > > This is not necessarily a bug or regression, but still fewer test suites > of stm-delay build at 3cfb12d8adac37e5565d66fd173e4648cc041e65 than > before at 5161609117c16cb7b29b2b8b1cd41e74341d4137 which is a suspicious > change to Stackage HEAD build report manager. > > There is a short description how it works here: https://github.com/ > tweag/stackage-head#how-it-works > > The logic that determines if a change in status of a package is suspicious > is here: > > https://github.com/tweag/stackage-head/blob/master/ > Stackage/HEAD/BuildDiff.hs#L61-L94 > > And the entire source code of the tool is available here: > > https://github.com/tweag/stackage-head > > As you can see, a package can either fail to build, or be unreachable (due > to failing dependencies) or it can succeed in which case we record number > of succeeding/failing test suites. > > Currently statistics look like this (with nightly-2018-04-05): > > * Failing packages: 5 > * Unreachable packages: 794 > * Packages that build: 897 > > The build takes about one hour. > > Circle CI logs are available here: > > https://circleci.com/gh/tweag/stackage-head > > I configured this to run 4 times per day for a start. I'll receive email > notifications on failures which I plan to convert to trac tickets manually > (for now.). > > I could not find a way to make Circle CI send emails to other people like > Ben, but I see there are various hook available, so we could use some of > them if desirable. > > Best, > > Mark > > > > _______________________________________________ > Ghc-devops-group mailing list > Ghc-devops-group at haskell.org > https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devops-group > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From manuel.chakravarty at tweag.io Thu Apr 12 05:27:43 2018 From: manuel.chakravarty at tweag.io (Manuel M T Chakravarty) Date: Thu, 12 Apr 2018 15:27:43 +1000 Subject: [GHC DevOps Group] Stackage HEAD, now working In-Reply-To: References: Message-ID: <3E2DC73F-429E-4ECC-A7A3-97A0858F7FFC@tweag.io> Thanks, Mark, great work! I would think, it should eventually move to the GHC org. Cheers, Manuel > 11.04.2018 18:45 Boespflug, Mathieu : > > Hi folks, > > Stackage HEAD is currently administered in the tweag GitHub org, due to practical constraints. But I'd encourage moving it over to the ghc or haskell orgs if the community finds this useful. > > Best, > > -- > Mathieu Boespflug > Founder at http://tweag.io . > > On 11 April 2018 at 10:40, Karpov, Mark > wrote: > Hello, > > Last year there was an initiative to execute Stackage nightly build plans with development versions of GHC and by comparing results that we obtain with different commits we could detect how they affect packages in our ecosystem. Thus we potentially can detect regressions and other suspicious changes. > > * The original blog post introducing the initiative: https://www.tweag.io/posts/2017-10-27-stackage-head.html > * A talk from HIW'17: https://www.youtube.com/watch?v=UAI-mplXUkY > > Recently we resumed working on this and actually I'm happy to announce that Stackage HEAD is functional now. In fact, a bit unexpectedly, it detected something today: > > https://ghc.haskell.org/trac/ghc/ticket/15022 > > This is not necessarily a bug or regression, but still fewer test suites of stm-delay build at 3cfb12d8adac37e5565d66fd173e4648cc041e65 than before at 5161609117c16cb7b29b2b8b1cd41e74341d4137 which is a suspicious change to Stackage HEAD build report manager. > > There is a short description how it works here: https://github.com/tweag/stackage-head#how-it-works > > The logic that determines if a change in status of a package is suspicious is here: > > https://github.com/tweag/stackage-head/blob/master/Stackage/HEAD/BuildDiff.hs#L61-L94 > > And the entire source code of the tool is available here: > > https://github.com/tweag/stackage-head > > As you can see, a package can either fail to build, or be unreachable (due to failing dependencies) or it can succeed in which case we record number of succeeding/failing test suites. > > Currently statistics look like this (with nightly-2018-04-05): > > * Failing packages: 5 > * Unreachable packages: 794 > * Packages that build: 897 > > The build takes about one hour. > > Circle CI logs are available here: > > https://circleci.com/gh/tweag/stackage-head > > I configured this to run 4 times per day for a start. I'll receive email notifications on failures which I plan to convert to trac tickets manually (for now.). > > I could not find a way to make Circle CI send emails to other people like Ben, but I see there are various hook available, so we could use some of them if desirable. > > Best, > > Mark > > > > _______________________________________________ > Ghc-devops-group mailing list > Ghc-devops-group at haskell.org > https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devops-group > > > _______________________________________________ > Ghc-devops-group mailing list > Ghc-devops-group at haskell.org > https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devops-group -------------- next part -------------- An HTML attachment was scrubbed... URL: From ben at well-typed.com Tue Apr 17 16:13:41 2018 From: ben at well-typed.com (Ben Gamari) Date: Tue, 17 Apr 2018 12:13:41 -0400 Subject: [GHC DevOps Group] U2Fs for GHC devops In-Reply-To: References: Message-ID: <87bmeherq1.fsf@smart-cactus.org> "Greg Steuck (Sh-toy-k)" writes: > I have obtained a few U2F devices from Google Advanced Protection program > . I sent one to Ben and I > am happy to send more to GHC devops. I used such devices for authenticating > to GitHub and Google, they make a good second factor. > Hi Greg! I have activated the device against a new account (bgamari.ghc at gmail.com) which I'll use for GHC devops matters (due to compatibility issues between GMail's U2F support and my mail configuration). Do you suppose you could remove access to the Google machines for bgamari at gmail.com and instead grant access to bgamari.ghc at gmail.com? Thanks again. Cheers, - Ben -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 487 bytes Desc: not available URL: