You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@continuum.apache.org by Brett Porter <br...@apache.org> on 2007/01/11 00:07:36 UTC
Trusting in our own dog food
Folks,
I'd like to turn off continuum_ci.sh and instead only use Continuum
itself to do CI for Continuum. Any objections?
- Brett
Re: Trusting in our own dog food
Posted by Arnaud HERITIER <ah...@gmail.com>.
It's good. I'm seeing it when I'm logged.
Arnaud
On 1/18/07, Brett Porter <br...@apache.org> wrote:
>
> try now
>
> On 18/01/2007, at 9:53 AM, Arnaud HERITIER wrote:
>
> > I just created an account and the list is also empty :-(
> >
> > Arnaud
> >
> > On 1/17/07, Brett Porter <br...@apache.org> wrote:
> >>
> >> was just pondering that myself.
> >>
> >> shouldn't the default be that guest is a user on all project groups
> >> and we remove it to be more restrictive?
> >>
> >> - Brett
> >>
> >> On 18/01/2007, at 8:05 AM, Arnaud HERITIER wrote:
> >>
> >> > Is it normal that the projects list is empty when we aren't logon ?
> >> >
> >> > http://maven.zones.apache.org:8080/continuum/groupSummary.action
> >> >
> >> > Arnaud
> >> >
> >> > On 1/17/07, Brett Porter <br...@apache.org> wrote:
> >> >>
> >> >> Ok, fair enough. I've left it on, and made it use a different
> >> local
> >> >> repository.
> >> >>
> >> >> I'd say once we release Continuum 1.1 and are happy it is stable
> >> >> enough to use, we can turn this off.
> >> >>
> >> >> On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
> >> >>
> >> >> > Brett Porter wrote:
> >> >> >> so... you're saying you don't trust our dog food? :)
> >> >> >
> >> >> > No, I'm saying it's there to verify the dog food. If there is no
> >> >> > discrepancies between what the cron is saying and the C
> >> instance is
> >> >> > saying, it's good. If there is an discrepancy it's not good.
> >> >> >
> >> >> > It will be more a tool to verification tool that a CI (but that
> >> >> > might be two sides of the same story :)
> >> >> >
> >> >> > --
> >> >> > Trygve
> >> >> >
> >> >> >> The only thing it tests differently is:
> >> >> >> - works by cron, whereas continuum might go down/hang/something
> >> >> >> else (which is something we should work on fixing if it does,
> >> >> >> rather than rely on ci.sh)
> >> >> >> - runs a reactor (can add that as a less frequent build
> >> execution
> >> >> >> in continuum too, though).
> >> >> >> So, I don't see any reason to keep it - wdyt?
> >> >> >> - Brett
> >> >> >> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
> >> >> >>> Brett Porter wrote:
> >> >> >>>> Folks,
> >> >> >>>> I'd like to turn off continuum_ci.sh and instead only use
> >> >> >>>> Continuum itself to do CI for Continuum. Any objections?
> >> >> >>>
> >> >> >>> I don't see why it should be turned off, but perhaps the
> >> >> >>> automatic notifications can be turned off or just send
> >> failures.
> >> >> >>> That way it would verify the product (it will in itself be an
> >> >> >>> integration test) because if the Continuum instance says that
> >> >> >>> something is failing, you should expect an email saying the
> >> same
> >> >> >>> right after. Or at least you can check the logs directory if
> >> >> >>> you're suspecting some other failure.
> >> >> >>>
> >> >> >>> --
> >> >> >>> Trygve
> >> >>
> >>
>
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
try now
On 18/01/2007, at 9:53 AM, Arnaud HERITIER wrote:
> I just created an account and the list is also empty :-(
>
> Arnaud
>
> On 1/17/07, Brett Porter <br...@apache.org> wrote:
>>
>> was just pondering that myself.
>>
>> shouldn't the default be that guest is a user on all project groups
>> and we remove it to be more restrictive?
>>
>> - Brett
>>
>> On 18/01/2007, at 8:05 AM, Arnaud HERITIER wrote:
>>
>> > Is it normal that the projects list is empty when we aren't logon ?
>> >
>> > http://maven.zones.apache.org:8080/continuum/groupSummary.action
>> >
>> > Arnaud
>> >
>> > On 1/17/07, Brett Porter <br...@apache.org> wrote:
>> >>
>> >> Ok, fair enough. I've left it on, and made it use a different
>> local
>> >> repository.
>> >>
>> >> I'd say once we release Continuum 1.1 and are happy it is stable
>> >> enough to use, we can turn this off.
>> >>
>> >> On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
>> >>
>> >> > Brett Porter wrote:
>> >> >> so... you're saying you don't trust our dog food? :)
>> >> >
>> >> > No, I'm saying it's there to verify the dog food. If there is no
>> >> > discrepancies between what the cron is saying and the C
>> instance is
>> >> > saying, it's good. If there is an discrepancy it's not good.
>> >> >
>> >> > It will be more a tool to verification tool that a CI (but that
>> >> > might be two sides of the same story :)
>> >> >
>> >> > --
>> >> > Trygve
>> >> >
>> >> >> The only thing it tests differently is:
>> >> >> - works by cron, whereas continuum might go down/hang/something
>> >> >> else (which is something we should work on fixing if it does,
>> >> >> rather than rely on ci.sh)
>> >> >> - runs a reactor (can add that as a less frequent build
>> execution
>> >> >> in continuum too, though).
>> >> >> So, I don't see any reason to keep it - wdyt?
>> >> >> - Brett
>> >> >> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
>> >> >>> Brett Porter wrote:
>> >> >>>> Folks,
>> >> >>>> I'd like to turn off continuum_ci.sh and instead only use
>> >> >>>> Continuum itself to do CI for Continuum. Any objections?
>> >> >>>
>> >> >>> I don't see why it should be turned off, but perhaps the
>> >> >>> automatic notifications can be turned off or just send
>> failures.
>> >> >>> That way it would verify the product (it will in itself be an
>> >> >>> integration test) because if the Continuum instance says that
>> >> >>> something is failing, you should expect an email saying the
>> same
>> >> >>> right after. Or at least you can check the logs directory if
>> >> >>> you're suspecting some other failure.
>> >> >>>
>> >> >>> --
>> >> >>> Trygve
>> >>
>>
Re: Trusting in our own dog food
Posted by Arnaud HERITIER <ah...@gmail.com>.
I just created an account and the list is also empty :-(
Arnaud
On 1/17/07, Brett Porter <br...@apache.org> wrote:
>
> was just pondering that myself.
>
> shouldn't the default be that guest is a user on all project groups
> and we remove it to be more restrictive?
>
> - Brett
>
> On 18/01/2007, at 8:05 AM, Arnaud HERITIER wrote:
>
> > Is it normal that the projects list is empty when we aren't logon ?
> >
> > http://maven.zones.apache.org:8080/continuum/groupSummary.action
> >
> > Arnaud
> >
> > On 1/17/07, Brett Porter <br...@apache.org> wrote:
> >>
> >> Ok, fair enough. I've left it on, and made it use a different local
> >> repository.
> >>
> >> I'd say once we release Continuum 1.1 and are happy it is stable
> >> enough to use, we can turn this off.
> >>
> >> On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
> >>
> >> > Brett Porter wrote:
> >> >> so... you're saying you don't trust our dog food? :)
> >> >
> >> > No, I'm saying it's there to verify the dog food. If there is no
> >> > discrepancies between what the cron is saying and the C instance is
> >> > saying, it's good. If there is an discrepancy it's not good.
> >> >
> >> > It will be more a tool to verification tool that a CI (but that
> >> > might be two sides of the same story :)
> >> >
> >> > --
> >> > Trygve
> >> >
> >> >> The only thing it tests differently is:
> >> >> - works by cron, whereas continuum might go down/hang/something
> >> >> else (which is something we should work on fixing if it does,
> >> >> rather than rely on ci.sh)
> >> >> - runs a reactor (can add that as a less frequent build execution
> >> >> in continuum too, though).
> >> >> So, I don't see any reason to keep it - wdyt?
> >> >> - Brett
> >> >> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
> >> >>> Brett Porter wrote:
> >> >>>> Folks,
> >> >>>> I'd like to turn off continuum_ci.sh and instead only use
> >> >>>> Continuum itself to do CI for Continuum. Any objections?
> >> >>>
> >> >>> I don't see why it should be turned off, but perhaps the
> >> >>> automatic notifications can be turned off or just send failures.
> >> >>> That way it would verify the product (it will in itself be an
> >> >>> integration test) because if the Continuum instance says that
> >> >>> something is failing, you should expect an email saying the same
> >> >>> right after. Or at least you can check the logs directory if
> >> >>> you're suspecting some other failure.
> >> >>>
> >> >>> --
> >> >>> Trygve
> >>
>
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
was just pondering that myself.
shouldn't the default be that guest is a user on all project groups
and we remove it to be more restrictive?
- Brett
On 18/01/2007, at 8:05 AM, Arnaud HERITIER wrote:
> Is it normal that the projects list is empty when we aren't logon ?
>
> http://maven.zones.apache.org:8080/continuum/groupSummary.action
>
> Arnaud
>
> On 1/17/07, Brett Porter <br...@apache.org> wrote:
>>
>> Ok, fair enough. I've left it on, and made it use a different local
>> repository.
>>
>> I'd say once we release Continuum 1.1 and are happy it is stable
>> enough to use, we can turn this off.
>>
>> On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
>>
>> > Brett Porter wrote:
>> >> so... you're saying you don't trust our dog food? :)
>> >
>> > No, I'm saying it's there to verify the dog food. If there is no
>> > discrepancies between what the cron is saying and the C instance is
>> > saying, it's good. If there is an discrepancy it's not good.
>> >
>> > It will be more a tool to verification tool that a CI (but that
>> > might be two sides of the same story :)
>> >
>> > --
>> > Trygve
>> >
>> >> The only thing it tests differently is:
>> >> - works by cron, whereas continuum might go down/hang/something
>> >> else (which is something we should work on fixing if it does,
>> >> rather than rely on ci.sh)
>> >> - runs a reactor (can add that as a less frequent build execution
>> >> in continuum too, though).
>> >> So, I don't see any reason to keep it - wdyt?
>> >> - Brett
>> >> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
>> >>> Brett Porter wrote:
>> >>>> Folks,
>> >>>> I'd like to turn off continuum_ci.sh and instead only use
>> >>>> Continuum itself to do CI for Continuum. Any objections?
>> >>>
>> >>> I don't see why it should be turned off, but perhaps the
>> >>> automatic notifications can be turned off or just send failures.
>> >>> That way it would verify the product (it will in itself be an
>> >>> integration test) because if the Continuum instance says that
>> >>> something is failing, you should expect an email saying the same
>> >>> right after. Or at least you can check the logs directory if
>> >>> you're suspecting some other failure.
>> >>>
>> >>> --
>> >>> Trygve
>>
Re: Trusting in our own dog food
Posted by Arnaud HERITIER <ah...@gmail.com>.
Is it normal that the projects list is empty when we aren't logon ?
http://maven.zones.apache.org:8080/continuum/groupSummary.action
Arnaud
On 1/17/07, Brett Porter <br...@apache.org> wrote:
>
> Ok, fair enough. I've left it on, and made it use a different local
> repository.
>
> I'd say once we release Continuum 1.1 and are happy it is stable
> enough to use, we can turn this off.
>
> On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
>
> > Brett Porter wrote:
> >> so... you're saying you don't trust our dog food? :)
> >
> > No, I'm saying it's there to verify the dog food. If there is no
> > discrepancies between what the cron is saying and the C instance is
> > saying, it's good. If there is an discrepancy it's not good.
> >
> > It will be more a tool to verification tool that a CI (but that
> > might be two sides of the same story :)
> >
> > --
> > Trygve
> >
> >> The only thing it tests differently is:
> >> - works by cron, whereas continuum might go down/hang/something
> >> else (which is something we should work on fixing if it does,
> >> rather than rely on ci.sh)
> >> - runs a reactor (can add that as a less frequent build execution
> >> in continuum too, though).
> >> So, I don't see any reason to keep it - wdyt?
> >> - Brett
> >> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
> >>> Brett Porter wrote:
> >>>> Folks,
> >>>> I'd like to turn off continuum_ci.sh and instead only use
> >>>> Continuum itself to do CI for Continuum. Any objections?
> >>>
> >>> I don't see why it should be turned off, but perhaps the
> >>> automatic notifications can be turned off or just send failures.
> >>> That way it would verify the product (it will in itself be an
> >>> integration test) because if the Continuum instance says that
> >>> something is failing, you should expect an email saying the same
> >>> right after. Or at least you can check the logs directory if
> >>> you're suspecting some other failure.
> >>>
> >>> --
> >>> Trygve
>
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
Ok, fair enough. I've left it on, and made it use a different local
repository.
I'd say once we release Continuum 1.1 and are happy it is stable
enough to use, we can turn this off.
On 15/01/2007, at 11:02 PM, Trygve Laugstøl wrote:
> Brett Porter wrote:
>> so... you're saying you don't trust our dog food? :)
>
> No, I'm saying it's there to verify the dog food. If there is no
> discrepancies between what the cron is saying and the C instance is
> saying, it's good. If there is an discrepancy it's not good.
>
> It will be more a tool to verification tool that a CI (but that
> might be two sides of the same story :)
>
> --
> Trygve
>
>> The only thing it tests differently is:
>> - works by cron, whereas continuum might go down/hang/something
>> else (which is something we should work on fixing if it does,
>> rather than rely on ci.sh)
>> - runs a reactor (can add that as a less frequent build execution
>> in continuum too, though).
>> So, I don't see any reason to keep it - wdyt?
>> - Brett
>> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
>>> Brett Porter wrote:
>>>> Folks,
>>>> I'd like to turn off continuum_ci.sh and instead only use
>>>> Continuum itself to do CI for Continuum. Any objections?
>>>
>>> I don't see why it should be turned off, but perhaps the
>>> automatic notifications can be turned off or just send failures.
>>> That way it would verify the product (it will in itself be an
>>> integration test) because if the Continuum instance says that
>>> something is failing, you should expect an email saying the same
>>> right after. Or at least you can check the logs directory if
>>> you're suspecting some other failure.
>>>
>>> --
>>> Trygve
Re: Trusting in our own dog food
Posted by Trygve Laugstøl <tr...@apache.org>.
Brett Porter wrote:
> so... you're saying you don't trust our dog food? :)
No, I'm saying it's there to verify the dog food. If there is no
discrepancies between what the cron is saying and the C instance is
saying, it's good. If there is an discrepancy it's not good.
It will be more a tool to verification tool that a CI (but that might be
two sides of the same story :)
--
Trygve
>
> The only thing it tests differently is:
> - works by cron, whereas continuum might go down/hang/something else
> (which is something we should work on fixing if it does, rather than
> rely on ci.sh)
> - runs a reactor (can add that as a less frequent build execution in
> continuum too, though).
>
> So, I don't see any reason to keep it - wdyt?
>
> - Brett
>
> On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
>
>> Brett Porter wrote:
>>> Folks,
>>> I'd like to turn off continuum_ci.sh and instead only use Continuum
>>> itself to do CI for Continuum. Any objections?
>>
>> I don't see why it should be turned off, but perhaps the automatic
>> notifications can be turned off or just send failures. That way it
>> would verify the product (it will in itself be an integration test)
>> because if the Continuum instance says that something is failing, you
>> should expect an email saying the same right after. Or at least you
>> can check the logs directory if you're suspecting some other failure.
>>
>> --
>> Trygve
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
so... you're saying you don't trust our dog food? :)
The only thing it tests differently is:
- works by cron, whereas continuum might go down/hang/something else
(which is something we should work on fixing if it does, rather than
rely on ci.sh)
- runs a reactor (can add that as a less frequent build execution in
continuum too, though).
So, I don't see any reason to keep it - wdyt?
- Brett
On 11/01/2007, at 7:57 PM, Trygve Laugstøl wrote:
> Brett Porter wrote:
>> Folks,
>> I'd like to turn off continuum_ci.sh and instead only use
>> Continuum itself to do CI for Continuum. Any objections?
>
> I don't see why it should be turned off, but perhaps the automatic
> notifications can be turned off or just send failures. That way it
> would verify the product (it will in itself be an integration test)
> because if the Continuum instance says that something is failing,
> you should expect an email saying the same right after. Or at least
> you can check the logs directory if you're suspecting some other
> failure.
>
> --
> Trygve
Re: Trusting in our own dog food
Posted by Trygve Laugstøl <tr...@apache.org>.
Brett Porter wrote:
> Folks,
>
> I'd like to turn off continuum_ci.sh and instead only use Continuum
> itself to do CI for Continuum. Any objections?
I don't see why it should be turned off, but perhaps the automatic
notifications can be turned off or just send failures. That way it would
verify the product (it will in itself be an integration test) because if
the Continuum instance says that something is failing, you should expect
an email saying the same right after. Or at least you can check the logs
directory if you're suspecting some other failure.
--
Trygve
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
yeah, it's subversion 1.1.4 (ouch!).
I'm going to look at upgrading!
On 11/01/2007, at 11:27 PM, Federico Yankelevich wrote:
>
> I read on svn changelog that SVN v1.4 increased a lot the speed for
> comparing
> local copy with repository.
> Maybe continuum is very slow in SVN update because it is using SVN
> 1.3 (both
> client and server needs to be updated)
>
> see http://subversion.tigris.org/svn_1.4_releasenotes.html
>
> just my 2 cents,
> Federico
>
>
>
> brettporter wrote:
>>
>> Yes, I have a script to automate installing the latest build (though
>> would need changes if continuum_ci was turned off).
>>
>> 1.1 is running very well thanks to some sleuthing by Wendy and quick
>> fixes from Emmanuel.
>>
>> My biggest concern is the scalability of polling. It currently takes
>> about 30 minutes to just run through all the required svn up commands
>> to detect if builds are needed for all the Maven projects.
>>
>> - Brett
>>
>> On 11/01/2007, at 10:26 AM, Arnaud HERITIER wrote:
>>
>>> good luck ;-)
>>> did you update the 2.1 snapshot ?
>>>
>>> Arnaud
>>>
>>> On 1/11/07, Brett Porter <br...@apache.org> wrote:
>>>>
>>>> Folks,
>>>>
>>>> I'd like to turn off continuum_ci.sh and instead only use Continuum
>>>> itself to do CI for Continuum. Any objections?
>>>>
>>>> - Brett
>>>>
>>>>
>>
>>
>
> --
> View this message in context: http://www.nabble.com/Trusting-in-our-
> own-dog-food-tf2955860.html#a8276485
> Sent from the Continuum - Dev mailing list archive at Nabble.com.
Re: Trusting in our own dog food
Posted by Federico Yankelevich <fy...@sobacosoftware.ch>.
I read on svn changelog that SVN v1.4 increased a lot the speed for comparing
local copy with repository.
Maybe continuum is very slow in SVN update because it is using SVN 1.3 (both
client and server needs to be updated)
see http://subversion.tigris.org/svn_1.4_releasenotes.html
just my 2 cents,
Federico
brettporter wrote:
>
> Yes, I have a script to automate installing the latest build (though
> would need changes if continuum_ci was turned off).
>
> 1.1 is running very well thanks to some sleuthing by Wendy and quick
> fixes from Emmanuel.
>
> My biggest concern is the scalability of polling. It currently takes
> about 30 minutes to just run through all the required svn up commands
> to detect if builds are needed for all the Maven projects.
>
> - Brett
>
> On 11/01/2007, at 10:26 AM, Arnaud HERITIER wrote:
>
>> good luck ;-)
>> did you update the 2.1 snapshot ?
>>
>> Arnaud
>>
>> On 1/11/07, Brett Porter <br...@apache.org> wrote:
>>>
>>> Folks,
>>>
>>> I'd like to turn off continuum_ci.sh and instead only use Continuum
>>> itself to do CI for Continuum. Any objections?
>>>
>>> - Brett
>>>
>>>
>
>
--
View this message in context: http://www.nabble.com/Trusting-in-our-own-dog-food-tf2955860.html#a8276485
Sent from the Continuum - Dev mailing list archive at Nabble.com.
Re: Trusting in our own dog food
Posted by Brett Porter <br...@apache.org>.
Yes, I have a script to automate installing the latest build (though
would need changes if continuum_ci was turned off).
1.1 is running very well thanks to some sleuthing by Wendy and quick
fixes from Emmanuel.
My biggest concern is the scalability of polling. It currently takes
about 30 minutes to just run through all the required svn up commands
to detect if builds are needed for all the Maven projects.
- Brett
On 11/01/2007, at 10:26 AM, Arnaud HERITIER wrote:
> good luck ;-)
> did you update the 2.1 snapshot ?
>
> Arnaud
>
> On 1/11/07, Brett Porter <br...@apache.org> wrote:
>>
>> Folks,
>>
>> I'd like to turn off continuum_ci.sh and instead only use Continuum
>> itself to do CI for Continuum. Any objections?
>>
>> - Brett
>>
>>
Re: Trusting in our own dog food
Posted by Arnaud HERITIER <ah...@gmail.com>.
good luck ;-)
did you update the 2.1 snapshot ?
Arnaud
On 1/11/07, Brett Porter <br...@apache.org> wrote:
>
> Folks,
>
> I'd like to turn off continuum_ci.sh and instead only use Continuum
> itself to do CI for Continuum. Any objections?
>
> - Brett
>
>