You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ctakes.apache.org by "Finan, Sean" <Se...@childrens.harvard.edu> on 2017/09/03 21:16:18 UTC

Re: Travis has one test failed [EXTERNAL]

Hi Andrey,

So you did set up a travis test for ctakes?  Cool.

In my opinion that test should be disabled for automated builds.  In addition to requiring credentials that should not be on a public system, it is basically a partial integration test that assures that a single workflow runs to completion.  As far as I know there is no check on generated output values, it just depends upon thrown exceptions or an absence of output files to reveal problems.  It also takes a long time to set up and run.  It produces a lot of output that has little to do with any actual testing.  There are names for these testing antipatterns:
the liar, the giant, the loudmouth, the slowpoke, and most of all, the secret catcher.

Please don't read into this; I am not trying to be mean or stating that the test doesn't have its place.  It can be a valuable sanity check for anybody making changes to the specific functionality that it covers. 
 People are welcome to disagree with me on this, but I think that this test is better done as-needed on an individual's system with their umls credentials.  This may create a "Local Hero" (another antipattern), but so what, you have to start somewhere.  A few people have slowly added unit tests to ctakes, and I hope that such valuable contributions continue.

That was probably more than you care to read; in summary I think that that particular test can be disabled for automation regression.  I can turn it off but will refrain for now in case others want to take another route.

Thanks again,
Sean
________________________________________
From: Andrey Kurdumov <ka...@googlemail.com>
Sent: Sunday, September 3, 2017 3:33 PM
To: cTakes developers list
Subject: Travis has one test failed [EXTERNAL]

The travis for cTakes has one test[1] failed due to missing UMLS license.
Maybe somebody could configure Travis to provide valid UMLS license, or
disable this test.

I would like to volunteer to help with this, but looks like this change is
for contributors only.

[1]
https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=

Re: Travis has one test failed [EXTERNAL]

Posted by Andrey Kurdumov <ka...@googlemail.com>.
Sorry, CI was on Jenkins which is part of Apache Infra. My bad that I
confuse this.
Somebody from contributors setup this 2 jobs.

One completely green - https://builds.apache.org/job/ctakes-trunk-package/
One has single test failed -
https://builds.apache.org/job/ctakes-trunk-compiletest/

Since this is cTakes test suite which running automatically, so it is not
for me to decide what is better :)


2017-09-04 3:16 GMT+06:00 Finan, Sean <Se...@childrens.harvard.edu>:

> Hi Andrey,
>
> So you did set up a travis test for ctakes?  Cool.
>
> In my opinion that test should be disabled for automated builds.  In
> addition to requiring credentials that should not be on a public system, it
> is basically a partial integration test that assures that a single workflow
> runs to completion.  As far as I know there is no check on generated output
> values, it just depends upon thrown exceptions or an absence of output
> files to reveal problems.  It also takes a long time to set up and run.  It
> produces a lot of output that has little to do with any actual testing.
> There are names for these testing antipatterns:
> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
> secret catcher.
>
> Please don't read into this; I am not trying to be mean or stating that
> the test doesn't have its place.  It can be a valuable sanity check for
> anybody making changes to the specific functionality that it covers.
>  People are welcome to disagree with me on this, but I think that this
> test is better done as-needed on an individual's system with their umls
> credentials.  This may create a "Local Hero" (another antipattern), but so
> what, you have to start somewhere.  A few people have slowly added unit
> tests to ctakes, and I hope that such valuable contributions continue.
>
> That was probably more than you care to read; in summary I think that that
> particular test can be disabled for automation regression.  I can turn it
> off but will refrain for now in case others want to take another route.
>
> Thanks again,
> Sean
> ________________________________________
> From: Andrey Kurdumov <ka...@googlemail.com>
> Sent: Sunday, September 3, 2017 3:33 PM
> To: cTakes developers list
> Subject: Travis has one test failed [EXTERNAL]
>
> The travis for cTakes has one test[1] failed due to missing UMLS license.
> Maybe somebody could configure Travis to provide valid UMLS license, or
> disable this test.
>
> I would like to volunteer to help with this, but looks like this change is
> for contributors only.
>
> [1]
> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.
> apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.
> apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.
> regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=
> qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=
> fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_
> z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-
> BCVF3i-mvJgUeZ4&e=
>

Re: Travis has one test failed [EXTERNAL] [SUSPICIOUS]

Posted by "Miller, Timothy" <Ti...@childrens.harvard.edu>.
Actually I think I haven't been getting these (from jenkins?), even when they are my fault. I checked in a change a few weeks ago and only noticed it caused a problem when I synced from another machine.
Tim

________________________________________
From: Finan, Sean <Se...@childrens.harvard.edu>
Sent: Sunday, September 3, 2017 5:59 PM
To: dev@ctakes.apache.org
Subject: Re: Travis has one test failed [EXTERNAL] [SUSPICIOUS]

Hi Steve,

I have gotten a couple that were my fault because I hadn't yet checked in all of the required changes / classes.  Terribly poor form on my part.  Those were the CC: emails.  Other than that I haven't gotten any.  I will double-check to make sure that I'm on notifications@ ... though if it fills my mailbox then maybe I don't want to  ...

Thanks,
Sean
________________________________________
From: Steven Bethard <st...@gmail.com>
Sent: Sunday, September 3, 2017 5:45 PM
To: dev@ctakes.apache.org
Subject: Re: Travis has one test failed [EXTERNAL]

Sean, aren't you subscribed to
notifications@ctakes.apache.org? I've been getting failed test messages
from the official cTakes build for weeks now. You've even been CC'd on some
of them. Maybe check your junk mail? I haven't said anything about these
emails because 1) I assumed everyone else was getting them and 2) I don't
have the time now to look into a fix.

Steve

On Sun, Sep 3, 2017, 14:16 Finan, Sean <Se...@childrens.harvard.edu>
wrote:

> Hi Andrey,
>
> So you did set up a travis test for ctakes?  Cool.
>
> In my opinion that test should be disabled for automated builds.  In
> addition to requiring credentials that should not be on a public system, it
> is basically a partial integration test that assures that a single workflow
> runs to completion.  As far as I know there is no check on generated output
> values, it just depends upon thrown exceptions or an absence of output
> files to reveal problems.  It also takes a long time to set up and run.  It
> produces a lot of output that has little to do with any actual testing.
> There are names for these testing antipatterns:
> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
> secret catcher.
>
> Please don't read into this; I am not trying to be mean or stating that
> the test doesn't have its place.  It can be a valuable sanity check for
> anybody making changes to the specific functionality that it covers.
>  People are welcome to disagree with me on this, but I think that this
> test is better done as-needed on an individual's system with their umls
> credentials.  This may create a "Local Hero" (another antipattern), but so
> what, you have to start somewhere.  A few people have slowly added unit
> tests to ctakes, and I hope that such valuable contributions continue.
>
> That was probably more than you care to read; in summary I think that that
> particular test can be disabled for automation regression.  I can turn it
> off but will refrain for now in case others want to take another route.
>
> Thanks again,
> Sean
> ________________________________________
> From: Andrey Kurdumov <ka...@googlemail.com>
> Sent: Sunday, September 3, 2017 3:33 PM
> To: cTakes developers list
> Subject: Travis has one test failed [EXTERNAL]
>
> The travis for cTakes has one test[1] failed due to missing UMLS license.
> Maybe somebody could configure Travis to provide valid UMLS license, or
> disable this test.
>
> I would like to volunteer to help with this, but looks like this change is
> for contributors only.
>
> [1]
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=
>

Re: Builds are now clean [was: Travis has one test failed [EXTERNAL]]

Posted by James Masanz <ma...@gmail.com>.
That's excellent!

My preference would be for you to continue to submit attachments to JIRA
issues.

James


On Sep 10, 2017 7:48 PM, "Hadrian Zbarcea" <hz...@gmail.com> wrote:

Hi all,

I had issues with the tests as well. We need clean tests if we were to do
any incremental refactoring. I added to patches to #334 [1] and #445 [2].
One of my comments for #334 is incorrect in that I thought there was one
more test failing, but the issue was not properly escaping my umls password
in the env var (not the best practice, but I'll leave this for another
time).

That said with the 2 patches above, now I get clean builds and tests with
java 8 on linux. I do not use windows, but I strongly suspect that it'll be
a clean build on windows as well.

Because of the delay in applying patches, I am not sure if best would be
for me to fork on github and start working on a branch and submit PRs or
continue with attachments to JIRA. Please let me know what the
committers/community preference is and I'll adapt.

Cheers,
Hadrian


[1] https://issues.apache.org/jira/browse/CTAKES-334
[2] https://issues.apache.org/jira/browse/CTAKES-445

On 09/03/2017 05:59 PM, Finan, Sean wrote:

> Hi Steve,
>
> I have gotten a couple that were my fault because I hadn't yet checked in
> all of the required changes / classes.  Terribly poor form on my part.
> Those were the CC: emails.  Other than that I haven't gotten any.  I will
> double-check to make sure that I'm on notifications@ ... though if it
> fills my mailbox then maybe I don't want to  ...
>
> Thanks,
> Sean
> ________________________________________
> From: Steven Bethard <st...@gmail.com>
> Sent: Sunday, September 3, 2017 5:45 PM
> To: dev@ctakes.apache.org
> Subject: Re: Travis has one test failed [EXTERNAL]
>
> Sean, aren't you subscribed to
> notifications@ctakes.apache.org? I've been getting failed test messages
> from the official cTakes build for weeks now. You've even been CC'd on some
> of them. Maybe check your junk mail? I haven't said anything about these
> emails because 1) I assumed everyone else was getting them and 2) I don't
> have the time now to look into a fix.
>
> Steve
>
> On Sun, Sep 3, 2017, 14:16 Finan, Sean <Se...@childrens.harvard.edu>
> wrote:
>
> Hi Andrey,
>>
>> So you did set up a travis test for ctakes?  Cool.
>>
>> In my opinion that test should be disabled for automated builds.  In
>> addition to requiring credentials that should not be on a public system,
>> it
>> is basically a partial integration test that assures that a single
>> workflow
>> runs to completion.  As far as I know there is no check on generated
>> output
>> values, it just depends upon thrown exceptions or an absence of output
>> files to reveal problems.  It also takes a long time to set up and run.
>> It
>> produces a lot of output that has little to do with any actual testing.
>> There are names for these testing antipatterns:
>> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
>> secret catcher.
>>
>> Please don't read into this; I am not trying to be mean or stating that
>> the test doesn't have its place.  It can be a valuable sanity check for
>> anybody making changes to the specific functionality that it covers.
>>   People are welcome to disagree with me on this, but I think that this
>> test is better done as-needed on an individual's system with their umls
>> credentials.  This may create a "Local Hero" (another antipattern), but so
>> what, you have to start somewhere.  A few people have slowly added unit
>> tests to ctakes, and I hope that such valuable contributions continue.
>>
>> That was probably more than you care to read; in summary I think that that
>> particular test can be disabled for automation regression.  I can turn it
>> off but will refrain for now in case others want to take another route.
>>
>> Thanks again,
>> Sean
>> ________________________________________
>> From: Andrey Kurdumov <ka...@googlemail.com>
>> Sent: Sunday, September 3, 2017 3:33 PM
>> To: cTakes developers list
>> Subject: Travis has one test failed [EXTERNAL]
>>
>> The travis for cTakes has one test[1] failed due to missing UMLS license.
>> Maybe somebody could configure Travis to provide valid UMLS license, or
>> disable this test.
>>
>> I would like to volunteer to help with this, but looks like this change is
>> for contributors only.
>>
>> [1]
>>
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.
>> apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.ap
>> ache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apac
>> he.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=
>> DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs6
>> 7GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10X
>> o1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BC
>> VF3i-mvJgUeZ4&e=
>>
>>

RE: Builds are now clean [was: Travis has one test failed [EXTERNAL]]

Posted by "Finan, Sean" <Se...@childrens.harvard.edu>.
Thanks Hadrian!
Sean

-----Original Message-----
From: Hadrian Zbarcea [mailto:hzbarcea@gmail.com] 
Sent: Sunday, September 10, 2017 8:48 PM
To: dev@ctakes.apache.org; Finan, Sean
Subject: Builds are now clean [was: Travis has one test failed [EXTERNAL]]

Hi all,

I had issues with the tests as well. We need clean tests if we were to do any incremental refactoring. I added to patches to #334 [1] and #445 [2]. One of my comments for #334 is incorrect in that I thought there was one more test failing, but the issue was not properly escaping my umls password in the env var (not the best practice, but I'll leave this for another time).

That said with the 2 patches above, now I get clean builds and tests with java 8 on linux. I do not use windows, but I strongly suspect that it'll be a clean build on windows as well.

Because of the delay in applying patches, I am not sure if best would be for me to fork on github and start working on a branch and submit PRs or continue with attachments to JIRA. Please let me know what the committers/community preference is and I'll adapt.

Cheers,
Hadrian


[1] https://urldefense.proofpoint.com/v2/url?u=https-3A__issues.apache.org_jira_browse_CTAKES-2D334&d=DwICaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=cIutJNIrNsLAO6pi3SEzysK-HWOHIPZUEee_slrLemQ&s=kWFA-JJHPLcE4CM1v_VTJZva2IYx9qoOgKTvf1CllnY&e=
[2] https://urldefense.proofpoint.com/v2/url?u=https-3A__issues.apache.org_jira_browse_CTAKES-2D445&d=DwICaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=cIutJNIrNsLAO6pi3SEzysK-HWOHIPZUEee_slrLemQ&s=St9xekBiqtNKT31KDDEnFtzwSzoUBfbNjkgr13KBTyg&e= 

On 09/03/2017 05:59 PM, Finan, Sean wrote:
> Hi Steve,
> 
> I have gotten a couple that were my fault because I hadn't yet checked in all of the required changes / classes.  Terribly poor form on my part.  Those were the CC: emails.  Other than that I haven't gotten any.  I will double-check to make sure that I'm on notifications@ ... though if it fills my mailbox then maybe I don't want to  ...
> 
> Thanks,
> Sean
> ________________________________________
> From: Steven Bethard <st...@gmail.com>
> Sent: Sunday, September 3, 2017 5:45 PM
> To: dev@ctakes.apache.org
> Subject: Re: Travis has one test failed [EXTERNAL]
> 
> Sean, aren't you subscribed to
> notifications@ctakes.apache.org? I've been getting failed test 
> messages from the official cTakes build for weeks now. You've even 
> been CC'd on some of them. Maybe check your junk mail? I haven't said 
> anything about these emails because 1) I assumed everyone else was 
> getting them and 2) I don't have the time now to look into a fix.
> 
> Steve
> 
> On Sun, Sep 3, 2017, 14:16 Finan, Sean 
> <Se...@childrens.harvard.edu>
> wrote:
> 
>> Hi Andrey,
>>
>> So you did set up a travis test for ctakes?  Cool.
>>
>> In my opinion that test should be disabled for automated builds.  In 
>> addition to requiring credentials that should not be on a public 
>> system, it is basically a partial integration test that assures that 
>> a single workflow runs to completion.  As far as I know there is no 
>> check on generated output values, it just depends upon thrown 
>> exceptions or an absence of output files to reveal problems.  It also 
>> takes a long time to set up and run.  It produces a lot of output that has little to do with any actual testing.
>> There are names for these testing antipatterns:
>> the liar, the giant, the loudmouth, the slowpoke, and most of all, 
>> the secret catcher.
>>
>> Please don't read into this; I am not trying to be mean or stating 
>> that the test doesn't have its place.  It can be a valuable sanity 
>> check for anybody making changes to the specific functionality that it covers.
>>   People are welcome to disagree with me on this, but I think that 
>> this test is better done as-needed on an individual's system with 
>> their umls credentials.  This may create a "Local Hero" (another 
>> antipattern), but so what, you have to start somewhere.  A few people 
>> have slowly added unit tests to ctakes, and I hope that such valuable contributions continue.
>>
>> That was probably more than you care to read; in summary I think that 
>> that particular test can be disabled for automation regression.  I 
>> can turn it off but will refrain for now in case others want to take another route.
>>
>> Thanks again,
>> Sean
>> ________________________________________
>> From: Andrey Kurdumov <ka...@googlemail.com>
>> Sent: Sunday, September 3, 2017 3:33 PM
>> To: cTakes developers list
>> Subject: Travis has one test failed [EXTERNAL]
>>
>> The travis for cTakes has one test[1] failed due to missing UMLS license.
>> Maybe somebody could configure Travis to provide valid UMLS license, 
>> or disable this test.
>>
>> I would like to volunteer to help with this, but looks like this 
>> change is for contributors only.
>>
>> [1]
>>
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.or
>> g_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=
>>

Builds are now clean [was: Travis has one test failed [EXTERNAL]]

Posted by Hadrian Zbarcea <hz...@gmail.com>.
Hi all,

I had issues with the tests as well. We need clean tests if we were to 
do any incremental refactoring. I added to patches to #334 [1] and #445 
[2]. One of my comments for #334 is incorrect in that I thought there 
was one more test failing, but the issue was not properly escaping my 
umls password in the env var (not the best practice, but I'll leave this 
for another time).

That said with the 2 patches above, now I get clean builds and tests 
with java 8 on linux. I do not use windows, but I strongly suspect that 
it'll be a clean build on windows as well.

Because of the delay in applying patches, I am not sure if best would be 
for me to fork on github and start working on a branch and submit PRs or 
continue with attachments to JIRA. Please let me know what the 
committers/community preference is and I'll adapt.

Cheers,
Hadrian


[1] https://issues.apache.org/jira/browse/CTAKES-334
[2] https://issues.apache.org/jira/browse/CTAKES-445

On 09/03/2017 05:59 PM, Finan, Sean wrote:
> Hi Steve,
> 
> I have gotten a couple that were my fault because I hadn't yet checked in all of the required changes / classes.  Terribly poor form on my part.  Those were the CC: emails.  Other than that I haven't gotten any.  I will double-check to make sure that I'm on notifications@ ... though if it fills my mailbox then maybe I don't want to  ...
> 
> Thanks,
> Sean
> ________________________________________
> From: Steven Bethard <st...@gmail.com>
> Sent: Sunday, September 3, 2017 5:45 PM
> To: dev@ctakes.apache.org
> Subject: Re: Travis has one test failed [EXTERNAL]
> 
> Sean, aren't you subscribed to
> notifications@ctakes.apache.org? I've been getting failed test messages
> from the official cTakes build for weeks now. You've even been CC'd on some
> of them. Maybe check your junk mail? I haven't said anything about these
> emails because 1) I assumed everyone else was getting them and 2) I don't
> have the time now to look into a fix.
> 
> Steve
> 
> On Sun, Sep 3, 2017, 14:16 Finan, Sean <Se...@childrens.harvard.edu>
> wrote:
> 
>> Hi Andrey,
>>
>> So you did set up a travis test for ctakes?  Cool.
>>
>> In my opinion that test should be disabled for automated builds.  In
>> addition to requiring credentials that should not be on a public system, it
>> is basically a partial integration test that assures that a single workflow
>> runs to completion.  As far as I know there is no check on generated output
>> values, it just depends upon thrown exceptions or an absence of output
>> files to reveal problems.  It also takes a long time to set up and run.  It
>> produces a lot of output that has little to do with any actual testing.
>> There are names for these testing antipatterns:
>> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
>> secret catcher.
>>
>> Please don't read into this; I am not trying to be mean or stating that
>> the test doesn't have its place.  It can be a valuable sanity check for
>> anybody making changes to the specific functionality that it covers.
>>   People are welcome to disagree with me on this, but I think that this
>> test is better done as-needed on an individual's system with their umls
>> credentials.  This may create a "Local Hero" (another antipattern), but so
>> what, you have to start somewhere.  A few people have slowly added unit
>> tests to ctakes, and I hope that such valuable contributions continue.
>>
>> That was probably more than you care to read; in summary I think that that
>> particular test can be disabled for automation regression.  I can turn it
>> off but will refrain for now in case others want to take another route.
>>
>> Thanks again,
>> Sean
>> ________________________________________
>> From: Andrey Kurdumov <ka...@googlemail.com>
>> Sent: Sunday, September 3, 2017 3:33 PM
>> To: cTakes developers list
>> Subject: Travis has one test failed [EXTERNAL]
>>
>> The travis for cTakes has one test[1] failed due to missing UMLS license.
>> Maybe somebody could configure Travis to provide valid UMLS license, or
>> disable this test.
>>
>> I would like to volunteer to help with this, but looks like this change is
>> for contributors only.
>>
>> [1]
>>
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=
>>

Re: Travis has one test failed [EXTERNAL]

Posted by "Finan, Sean" <Se...@childrens.harvard.edu>.
Hi Steve,

I have gotten a couple that were my fault because I hadn't yet checked in all of the required changes / classes.  Terribly poor form on my part.  Those were the CC: emails.  Other than that I haven't gotten any.  I will double-check to make sure that I'm on notifications@ ... though if it fills my mailbox then maybe I don't want to  ...

Thanks,
Sean
________________________________________
From: Steven Bethard <st...@gmail.com>
Sent: Sunday, September 3, 2017 5:45 PM
To: dev@ctakes.apache.org
Subject: Re: Travis has one test failed [EXTERNAL]

Sean, aren't you subscribed to
notifications@ctakes.apache.org? I've been getting failed test messages
from the official cTakes build for weeks now. You've even been CC'd on some
of them. Maybe check your junk mail? I haven't said anything about these
emails because 1) I assumed everyone else was getting them and 2) I don't
have the time now to look into a fix.

Steve

On Sun, Sep 3, 2017, 14:16 Finan, Sean <Se...@childrens.harvard.edu>
wrote:

> Hi Andrey,
>
> So you did set up a travis test for ctakes?  Cool.
>
> In my opinion that test should be disabled for automated builds.  In
> addition to requiring credentials that should not be on a public system, it
> is basically a partial integration test that assures that a single workflow
> runs to completion.  As far as I know there is no check on generated output
> values, it just depends upon thrown exceptions or an absence of output
> files to reveal problems.  It also takes a long time to set up and run.  It
> produces a lot of output that has little to do with any actual testing.
> There are names for these testing antipatterns:
> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
> secret catcher.
>
> Please don't read into this; I am not trying to be mean or stating that
> the test doesn't have its place.  It can be a valuable sanity check for
> anybody making changes to the specific functionality that it covers.
>  People are welcome to disagree with me on this, but I think that this
> test is better done as-needed on an individual's system with their umls
> credentials.  This may create a "Local Hero" (another antipattern), but so
> what, you have to start somewhere.  A few people have slowly added unit
> tests to ctakes, and I hope that such valuable contributions continue.
>
> That was probably more than you care to read; in summary I think that that
> particular test can be disabled for automation regression.  I can turn it
> off but will refrain for now in case others want to take another route.
>
> Thanks again,
> Sean
> ________________________________________
> From: Andrey Kurdumov <ka...@googlemail.com>
> Sent: Sunday, September 3, 2017 3:33 PM
> To: cTakes developers list
> Subject: Travis has one test failed [EXTERNAL]
>
> The travis for cTakes has one test[1] failed due to missing UMLS license.
> Maybe somebody could configure Travis to provide valid UMLS license, or
> disable this test.
>
> I would like to volunteer to help with this, but looks like this change is
> for contributors only.
>
> [1]
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=
>

Re: Travis has one test failed [EXTERNAL]

Posted by Steven Bethard <st...@gmail.com>.
Sean, aren't you subscribed to
notifications@ctakes.apache.org? I've been getting failed test messages
from the official cTakes build for weeks now. You've even been CC'd on some
of them. Maybe check your junk mail? I haven't said anything about these
emails because 1) I assumed everyone else was getting them and 2) I don't
have the time now to look into a fix.

Steve

On Sun, Sep 3, 2017, 14:16 Finan, Sean <Se...@childrens.harvard.edu>
wrote:

> Hi Andrey,
>
> So you did set up a travis test for ctakes?  Cool.
>
> In my opinion that test should be disabled for automated builds.  In
> addition to requiring credentials that should not be on a public system, it
> is basically a partial integration test that assures that a single workflow
> runs to completion.  As far as I know there is no check on generated output
> values, it just depends upon thrown exceptions or an absence of output
> files to reveal problems.  It also takes a long time to set up and run.  It
> produces a lot of output that has little to do with any actual testing.
> There are names for these testing antipatterns:
> the liar, the giant, the loudmouth, the slowpoke, and most of all, the
> secret catcher.
>
> Please don't read into this; I am not trying to be mean or stating that
> the test doesn't have its place.  It can be a valuable sanity check for
> anybody making changes to the specific functionality that it covers.
>  People are welcome to disagree with me on this, but I think that this
> test is better done as-needed on an individual's system with their umls
> credentials.  This may create a "Local Hero" (another antipattern), but so
> what, you have to start somewhere.  A few people have slowly added unit
> tests to ctakes, and I hope that such valuable contributions continue.
>
> That was probably more than you care to read; in summary I think that that
> particular test can be disabled for automation regression.  I can turn it
> off but will refrain for now in case others want to take another route.
>
> Thanks again,
> Sean
> ________________________________________
> From: Andrey Kurdumov <ka...@googlemail.com>
> Sent: Sunday, September 3, 2017 3:33 PM
> To: cTakes developers list
> Subject: Travis has one test failed [EXTERNAL]
>
> The travis for cTakes has one test[1] failed due to missing UMLS license.
> Maybe somebody could configure Travis to provide valid UMLS license, or
> disable this test.
>
> I would like to volunteer to help with this, but looks like this change is
> for contributors only.
>
> [1]
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__builds.apache.org_job_ctakes-2Dtrunk-2Dcompiletest_lastBuild_org.apache.ctakes-24ctakes-2Dregression-2Dtest_testReport_org.apache.ctakes.regression.test_RegressionPipelineTest_testCPE_&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=4yTefwCxCfgP_z10Xo1oqK-A1NyZxXXt-2AvIvd6D7A&s=P3h16k70ouJKAOji8P5Wc8NRuLs-BCVF3i-mvJgUeZ4&e=
>