Disabling Travis?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Disabling Travis?

Joachim Breitner-2
Hi,

I wasn’t looking for a while I notice that the travis CI build of GHC
has not been succeeding for two months, partly because build time
exceeded the already much extended limits we have, partly because
support for booting with 7.10 was dropped, but the .travis file was not
used.

A CI system that is not normally succeeding is pretty pointless, and
will only confuse people who submit pull requests over GitHub who see
failures.

So if noone complains now I will soon disable building on Travis.

Greetings,
Joachim


--
Joachim “nomeata” Breitner
  [hidden email]
  https://www.joachim-breitner.de/

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Richard Eisenberg-4
Indeed I wrote and then deleted an email to you asking to update the Travis script to work again. I deleted it because I realized that fixing this myself was easy. So, at least one person cares about Travis. :)

Why do I care? Because I find it easier to validate with Travis than with Phab, for these reasons:

- I don't have to go through the process of creating a new diff -- no arc, no diff #, no reviewers that I don't care for (as I just want validation).

- Phab does advanced Sorcerer's-apprentice-level wizardry in applying my patch. If it's not rebased to a current master, Strange Things happen. And sometimes, I'm not ready to rebase (as I just want validation).

One might ask these questions at this point:

Q: Shouldn't I *always* create a diff for a change I'm making?

A: Perhaps I should. But sometimes a fix is very straightforward and I simply don't have the cycles to go through a review process for every patch. As any committer has learned, SPJ reviews every commit that goes by anyway, and it seems more agile just to commit what seem like simple patches and then commit tweaks.

Q: Why don't I just validate locally?

A: Because I'm on a Mac laptop. There are frequently tests that fail on Mac that aren't my fault, and sometimes I'm about to need to move the laptop -- and I don't want this to interrupt validation.

The big minus to Travis, as I see it, are that only committers can use it. (A forked repo just doesn't work because of the way that submodules are checked out, IIRC.) This minus doesn't affect me, however.

Thanks,
Richard

> On Aug 20, 2017, at 12:11 PM, Joachim Breitner <[hidden email]> wrote:
>
> Hi,
>
> I wasn’t looking for a while I notice that the travis CI build of GHC
> has not been succeeding for two months, partly because build time
> exceeded the already much extended limits we have, partly because
> support for booting with 7.10 was dropped, but the .travis file was not
> used.
>
> A CI system that is not normally succeeding is pretty pointless, and
> will only confuse people who submit pull requests over GitHub who see
> failures.
>
> So if noone complains now I will soon disable building on Travis.
>
> Greetings,
> Joachim
>
>
> --
> Joachim “nomeata” Breitner
>  [hidden email]
>  https://www.joachim-breitner.de/
> _______________________________________________
> ghc-devs mailing list
> [hidden email]
> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Joachim Breitner-2
Hi,

Am Sonntag, den 20.08.2017, 23:55 -0400 schrieb Richard Eisenberg:
> The big minus to Travis, as I see it, are that only committers can
> use it. (A forked repo just doesn't work because of the way that
> submodules are checked out, IIRC.) This minus doesn't affect me,
> however.

I think the biggest minus is that Travis can either be configured to
send mails to one specific address (as it was initially, when I
received all the mail and could check whether I want to notify the
author), or to the author of the patch, but not both. Currently it is
sent to notify the author, but people seem to simply ignore the mails. 

(Or maybe even only authors who also registered on travis get them?)

So if there is demand to keep using Travis, maybe we have to go back to
sending notification mails to one person who oversees this service?

Joachim



--
Joachim Breitner
  [hidden email]
  http://www.joachim-breitner.de/

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Richard Eisenberg-4
I have not been getting emails from Travis, despite using it for other projects (and thus being registered). So perhaps something is going wrong there. Regardless, it's useful for me, even without the emails.

Another nice thing about Travis: I can get a quick check of the history of failures, even in DEBUG mode.

Richard

> On Aug 21, 2017, at 3:54 AM, Joachim Breitner <[hidden email]> wrote:
>
> Hi,
>
> Am Sonntag, den 20.08.2017, 23:55 -0400 schrieb Richard Eisenberg:
>> The big minus to Travis, as I see it, are that only committers can
>> use it. (A forked repo just doesn't work because of the way that
>> submodules are checked out, IIRC.) This minus doesn't affect me,
>> however.
>
> I think the biggest minus is that Travis can either be configured to
> send mails to one specific address (as it was initially, when I
> received all the mail and could check whether I want to notify the
> author), or to the author of the patch, but not both. Currently it is
> sent to notify the author, but people seem to simply ignore the mails.
>
> (Or maybe even only authors who also registered on travis get them?)
>
> So if there is demand to keep using Travis, maybe we have to go back to
> sending notification mails to one person who oversees this service?
>
> Joachim
>
>
>
> --
> Joachim Breitner
>  [hidden email]
>  http://www.joachim-breitner.de/
> _______________________________________________
> ghc-devs mailing list
> [hidden email]
> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Joachim Breitner-2
Hi Richard,

I saw you tried to get Travis working again. I also gave it a quick
shot (in wip/travis2), but hit time limits again. Did you have more
luck?

Greetings,
Joachim

Am Montag, den 21.08.2017, 23:17 -0400 schrieb Richard Eisenberg:

> I have not been getting emails from Travis, despite using it for
> other projects (and thus being registered). So perhaps something is
> going wrong there. Regardless, it's useful for me, even without the
> emails.
>
> Another nice thing about Travis: I can get a quick check of the
> history of failures, even in DEBUG mode.
>
> Richard
>
> > On Aug 21, 2017, at 3:54 AM, Joachim Breitner <mail@joachim-breitne
> > r.de> wrote:
> >
> > Hi,
> >
> > Am Sonntag, den 20.08.2017, 23:55 -0400 schrieb Richard Eisenberg:
> > > The big minus to Travis, as I see it, are that only committers
> > > can
> > > use it. (A forked repo just doesn't work because of the way that
> > > submodules are checked out, IIRC.) This minus doesn't affect me,
> > > however.
> >
> > I think the biggest minus is that Travis can either be configured
> > to
> > send mails to one specific address (as it was initially, when I
> > received all the mail and could check whether I want to notify the
> > author), or to the author of the patch, but not both. Currently it
> > is
> > sent to notify the author, but people seem to simply ignore the
> > mails. 
> >
> > (Or maybe even only authors who also registered on travis get
> > them?)
> >
> > So if there is demand to keep using Travis, maybe we have to go
> > back to
> > sending notification mails to one person who oversees this service?
> >
> > Joachim
> >
> >
> >
> > -- 
> > Joachim Breitner
> >  [hidden email]
> >  http://www.joachim-breitner.de/
> > _______________________________________________
> > ghc-devs mailing list
> > [hidden email]
> > http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
>
> _______________________________________________
> ghc-devs mailing list
> [hidden email]
> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
--
Joachim “nomeata” Breitner
  [hidden email]
  https://www.joachim-breitner.de/

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Richard Eisenberg-4
No -- I hit the limits, too. :(

One of my challenges in getting patches in is that I have no reliable way of validating. Validating locally is possible, but it gets false negatives (Mac validation usually isn't clean) and is inconvenient. I will be working with my school's sysadmin to get a Linux server working to do this, but not everyone has that possibility.

Richard

> On Sep 3, 2017, at 9:55 AM, Joachim Breitner <[hidden email]> wrote:
>
> Hi Richard,
>
> I saw you tried to get Travis working again. I also gave it a quick
> shot (in wip/travis2), but hit time limits again. Did you have more
> luck?
>
> Greetings,
> Joachim
>
> Am Montag, den 21.08.2017, 23:17 -0400 schrieb Richard Eisenberg:
>> I have not been getting emails from Travis, despite using it for
>> other projects (and thus being registered). So perhaps something is
>> going wrong there. Regardless, it's useful for me, even without the
>> emails.
>>
>> Another nice thing about Travis: I can get a quick check of the
>> history of failures, even in DEBUG mode.
>>
>> Richard
>>
>>> On Aug 21, 2017, at 3:54 AM, Joachim Breitner <mail@joachim-breitne
>>> r.de> wrote:
>>>
>>> Hi,
>>>
>>> Am Sonntag, den 20.08.2017, 23:55 -0400 schrieb Richard Eisenberg:
>>>> The big minus to Travis, as I see it, are that only committers
>>>> can
>>>> use it. (A forked repo just doesn't work because of the way that
>>>> submodules are checked out, IIRC.) This minus doesn't affect me,
>>>> however.
>>>
>>> I think the biggest minus is that Travis can either be configured
>>> to
>>> send mails to one specific address (as it was initially, when I
>>> received all the mail and could check whether I want to notify the
>>> author), or to the author of the patch, but not both. Currently it
>>> is
>>> sent to notify the author, but people seem to simply ignore the
>>> mails.
>>>
>>> (Or maybe even only authors who also registered on travis get
>>> them?)
>>>
>>> So if there is demand to keep using Travis, maybe we have to go
>>> back to
>>> sending notification mails to one person who oversees this service?
>>>
>>> Joachim
>>>
>>>
>>>
>>> --
>>> Joachim Breitner
>>>  [hidden email]
>>>  http://www.joachim-breitner.de/
>>> _______________________________________________
>>> ghc-devs mailing list
>>> [hidden email]
>>> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
>>
>> _______________________________________________
>> ghc-devs mailing list
>> [hidden email]
>> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
> --
> Joachim “nomeata” Breitner
>  [hidden email]
>  https://www.joachim-breitner.de/
> _______________________________________________
> ghc-devs mailing list
> [hidden email]
> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Reply | Threaded
Open this post in threaded view
|

Re: Disabling Travis?

Boespflug, Mathieu
It took me no more than a couple hours to get this working, but using
CircleCI, for our fork of GHC. I started from Joachim's TravisCI
script.

https://circleci.com/gh/tweag/ghc/tree/circleci

It would be trivial to activate this for github.com/ghc/ghc as well.

A few notes:
- It runs ./validate --fast in 40 minutes.
- CircleCI has OS X support as well. I think we should just migrate to
using CircleCI for OS X testing instead of the custom drones, one or
all of which are currently down.
- CircleCI graciously agreed to running on one of the beefy AWS node
types, called c4.xlarge (8 cores). On the standard node type (2
cores), validate takes just over an hour to run. It would be great if
./validate could scale better to more cores.
_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Reply | Threaded
Open this post in threaded view
|

CircleCI (Was: Disable Travis?)

Joachim Breitner-2
Hi,

Am Donnerstag, den 21.09.2017, 14:25 +0200 schrieb Boespflug, Mathieu:

> It took me no more than a couple hours to get this working, but using
> CircleCI, for our fork of GHC. I started from Joachim's TravisCI
> script.
>
> https://circleci.com/gh/tweag/ghc/tree/circleci
>
> It would be trivial to activate this for github.com/ghc/ghc as well.
>
> A few notes:
> - It runs ./validate --fast in 40 minutes.
> - CircleCI has OS X support as well. I think we should just migrate
> to
> using CircleCI for OS X testing instead of the custom drones, one or
> all of which are currently down.
> - CircleCI graciously agreed to running on one of the beefy AWS node
> types, called c4.xlarge (8 cores). On the standard node type (2
> cores), validate takes just over an hour to run. It would be great if
> ./validate could scale better to more cores.
nice! Yes, let’s do this. More CI never hurts (if someone keeps an eye
on it and fixes breakage that is not due to the code).

Can you configure circleci to mail both the committeer and a specific
person (e.g. you, or me) on every failed committ?

I enabled it now for ghc/ghc, but it says

> Configurable resource class is not enabled in your project. Please
> contact your CSM person or our support team to whitelist your
> project.

Can you do that?

Greetings,
Joachim


--
Joachim Breitner
  [hidden email]
  http://www.joachim-breitner.de/

_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

signature.asc (849 bytes) Download Attachment