Re: [squeak-dev] minor errors in tests during socket timeouts

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

Re: [squeak-dev] minor errors in tests during socket timeouts

fniephaus
 


On Mon, 31 Dec 2018 at 12:07 am, Ben Coman <[hidden email]> wrote:


On Mon, 31 Dec 2018 at 02:22, Levente Uzonyi <[hidden email]> wrote:
Thanks Ben. I don't know what image version the CI uses,

The log provides the following info...
  Squeak5.1
  latest update: #16551
  Current Change Set: Unnamed1
  Image format 6521 (32 bit)
 
but Nebraska-ul.54 in the Trunk should fix the issue with
StringSocketTestCase.

Good to know, thanks Levente. Would Squeak 5.2 cover Nebraska-ul.54 ?
It seems SQUEAK_VERSION could be updated here...

However SmalltalkCI README indicates it doesn't support 32-bit of latest images for Squeak or Pharo...
but is that info current? These files seem to indicate support...

Squeak[64]-trunk means Squeak-trunk (32bit) and Squeak64-trunk are supported. Eventually, we want to introduce Squeak32-trunk as well and change the default to 64bit. 
The same rule applies for Pharo.
Maybe open an issue against smalltalkCI, so we don't forget to get rid of this confusion?




@vm-dev... 
I find the specification of the test image too implicit here...
For example, its hard to tell at a glance whether lowcode testing is enabled/disabled.
Could the test image be explicitly defined in travis.yml ? (I'll submit a concept PR for review)

Please do, happy to review a PR. The current version is still the very first version, I'm sure it can use some improvements :)

Fabio


cheers -ben


Levente

On Sun, 30 Dec 2018, Ben Coman wrote:

> An Opensmalltalk-vm Travis build job 
>     https://travis-ci.org/OpenSmalltalk/opensmalltalk-vm/jobs/473535608
> looks like it failed with a socket error, probably external network.  
> A bundle of failures look like this... 
>
>   SocketStreamTest
>
>   7041f6_0fb1
>    ✗ #testClassComment (10003ms)
>   TestFailure: Test timed out
>   Process>>pvtSignal:list:
>   [] in DelayWaitTimeout>>wait
>   <snip>
>
> which is okay, but the following three with similar >10000ms timeout
> did not produce "TestFailure: Test timed out " 
> but instead "MessageNotUnderstood" errors.
>
>
>   StringSocketTestCase
>
>   c43ad5_0fb1
>    ✗ #testBasics (10001ms)
>   MessageNotUnderstood: UndefinedObject>>destroy
>   UndefinedObject(Object)>>doesNotUnderstand: #destroy
>   StringSocketTestCase>>tearDown
>
>   e360b0_0fb1
>    ✗ #testBogusInput1 (10000ms)
>   MessageNotUnderstood: UndefinedObject>>destroy
>   UndefinedObject(Object)>>doesNotUnderstand: #destroy
>   StringSocketTestCase>>tearDown 
>
>   e4f711_0fb1
>    ✗ #testBogusInput2 (10002ms)
>   MessageNotUnderstood: UndefinedObject>>destroy
>   UndefinedObject(Object)>>doesNotUnderstand: #destroy
>   StringSocketTestCase>>tearDown
>
>
> I'm not currently running Squeak, so I'm just reporting was I see in the CI log
> in case anyone was interested in touching-up such a corner case. 
> Is it preferable to report on Mantis? Or not important :)
>
> cheers -ben
>
>

Reply | Threaded
Open this post in threaded view
|

Re: [squeak-dev] minor errors in tests during socket timeouts

Ben Coman
 
On Mon, 31 Dec 2018 at 07:20, Fabio Niephaus <[hidden email]> wrote:

>
> On Mon, 31 Dec 2018 at 12:07 am, Ben Coman <[hidden email]> wrote:
>>
>> However SmalltalkCI README indicates it doesn't support 32-bit of latest images for Squeak or Pharo...
>>    https://github.com/hpi-swa/smalltalkCI#list-of-supported-images
>> but is that info current? These files seem to indicate support...
>>    https://github.com/hpi-swa/smalltalkCI/blob/master/squeak/run.sh
>>    https://github.com/hpi-swa/smalltalkCI/blob/master/pharo/run.sh
>
>
> Squeak[64]-trunk means Squeak-trunk (32bit) and Squeak64-trunk are supported. Eventually, we want to introduce Squeak32-trunk as well and change the default to 64bit.
> The same rule applies for Pharo.
> Maybe open an issue against smalltalkCI, so we don't forget to get rid of this confusion?

https://github.com/hpi-swa/smalltalkCI/issues/407


>> @vm-dev...
>> I find the specification of the test image too implicit here...
>>    https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/tests/smalltalkCI.sh
>> For example, its hard to tell at a glance whether lowcode testing is enabled/disabled.
>> Could the test image be explicitly defined in travis.yml ? (I'll submit a concept PR for review)
>
>
> Please do, happy to review a PR. The current version is still the very first version, I'm sure it can use some improvements :)

cool.

btw, are we indeed not testing VMs on Windows?
  https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/.appveyor.yml#L123

cheers -ben
Reply | Threaded
Open this post in threaded view
|

explicitly specify CI test images (was Re: minor errors in tests during socket timeouts)

Ben Coman
 
On Mon, 31 Dec 2018 at 07:45, Ben Coman <[hidden email]> wrote:
On Mon, 31 Dec 2018 at 07:20, Fabio Niephaus <[hidden email]> wrote:
>
> On Mon, 31 Dec 2018 at 12:07 am, Ben Coman <[hidden email]> wrote:
>>
>> I find the specification of the test image too implicit here...
>>    https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/tests/smalltalkCI.sh
>> For example, its hard to tell at a glance whether lowcode testing is enabled/disabled.
>> Could the test image be explicitly defined in travis.yml ? (I'll submit a concept PR for review)
>
>
> Please do, happy to review a PR. The current version is still the very first version, I'm sure it can use some improvements :)

Here are my proposed changes...

Here it is a sample log of the first job...

So the goal was that you can hover over the environment variables to easily review the test configuration here... 

I'm not sure how the newspeak builds should be handled differently,
since they use an alternate test script from here...

cheers -ben 
 

btw, are we indeed not testing VMs on Windows?
  https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/.appveyor.yml#L123
Reply | Threaded
Open this post in threaded view
|

Re: [squeak-dev] minor errors in tests during socket timeouts

fniephaus
In reply to this post by Ben Coman
 

On Mon, Dec 31, 2018 at 12:46 AM Ben Coman <[hidden email]> wrote:
 
On Mon, 31 Dec 2018 at 07:20, Fabio Niephaus <[hidden email]> wrote:
>
> On Mon, 31 Dec 2018 at 12:07 am, Ben Coman <[hidden email]> wrote:
>>
>> However SmalltalkCI README indicates it doesn't support 32-bit of latest images for Squeak or Pharo...
>>    https://github.com/hpi-swa/smalltalkCI#list-of-supported-images
>> but is that info current? These files seem to indicate support...
>>    https://github.com/hpi-swa/smalltalkCI/blob/master/squeak/run.sh
>>    https://github.com/hpi-swa/smalltalkCI/blob/master/pharo/run.sh
>
>
> Squeak[64]-trunk means Squeak-trunk (32bit) and Squeak64-trunk are supported. Eventually, we want to introduce Squeak32-trunk as well and change the default to 64bit.
> The same rule applies for Pharo.
> Maybe open an issue against smalltalkCI, so we don't forget to get rid of this confusion?

https://github.com/hpi-swa/smalltalkCI/issues/407

Fixed!
 



>> @vm-dev...
>> I find the specification of the test image too implicit here...
>>    https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/tests/smalltalkCI.sh
>> For example, its hard to tell at a glance whether lowcode testing is enabled/disabled.
>> Could the test image be explicitly defined in travis.yml ? (I'll submit a concept PR for review)
>
>
> Please do, happy to review a PR. The current version is still the very first version, I'm sure it can use some improvements :)

cool.

btw, are we indeed not testing VMs on Windows?
  https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/.appveyor.yml#L123

I'm afraid you're right. If you could set that up in a similar, that'd be great. smalltalkCI does support AppVeyor:

Fabio
 


cheers -ben