Test failure: System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests/ValidateAgainstBlobs(obj: CookieContainer { Capacity = 10, Count = 1, MaxCookieSize = 1024, PerDomainCapacity = 5 }, blobs: [...

[ghost]

Opened on behalf of @danmosemsft

The test System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests/ValidateAgainstBlobs(obj: CookieContainer { Capacity = 10, Count = 1, MaxCookieSize = 1024, PerDomainCapacity = 5 }, blobs: [... has failed.

The stored blob for type System.Net.CookieContainer is outdated and needs to be updated.
Stored blob: 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
Generated runtime blob: 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
Expected: True
Actual: False

    Stack Trace:

       at System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests.ValidateAgainstBlobs(Object obj, String[] blobs) in /Users/buildagent/agent/_work/45/s/corefx/src/System.Runtime.Serialization.Formatters/tests/BinaryFormatterTests.cs:line 62

Build : Master - 20170626.01 (Core Tests)
Failing configurations:

• OSX.1013.Amd64-x64
  • Release
• OSX.1012.Amd64-x64
  • Release

[danmoseley]

@joperezr Mac issue I mentioned

I decoded and I see
<<m_SortVersionculture ? System.Globalization.SortVersion???en-US
>>m_SortVersionculture ? System.Globalization.SortVersion

@ViktorHofer says The good thing: The serialization payload isn’t missing any fields, all names are in it. Only the value at the end of the payload is missing: en-US. After a quick look it seems like en-US is the value of the field m_name in CompareInfo. I don’t know why it is missing in the payload only for OSX. Was this the first time it failed? Maybe this has to do with some production code differences of CompareInfo on OSX?

[danmoseley]

Ask @ViktorHofer if you need help

[joperezr]

I can't repro this on my Mac, I have 10.12.5 (16F73) version installed and running latest master netcoreapp tests on Release configuration.

[joperezr]

I checked the RuntimeInformation of the test that @danmosemsft added, and here are the differences:

RuntimeInfo of my machine (test passes)

OS=Darwin 16.6.0 Darwin Kernel Version 16.6.0: Fri Apr 14 16:21:16 PDT 2017; root:xnu-3789.60.24~6/RELEASE_X86_64 OSVer=Unix 16.6.0.0 OSArch=X64 Arch=X64 Framework=.NET Core 4.6.00001.0

Runtime Info on Helix (tests do not pass)

OS=Darwin 16.4.0 Darwin Kernel Version 16.4.0: Thu Dec 22 22:53:21 PST 2016; root:xnu-3789.41.3~3/RELEASE_X86_64 OSVer=Unix 16.4.0.0 OSArch=X64 Arch=X64 Framework=.NET Core 4.6.00001.0

[danmoseley]

@ViktorHofer there are similar errors on Windows 10 and Fedora, in the 2.0 branch (not nesc same cause)

https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170704.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20CookieContainer%20%7B%20Capacity%20=%2010,%20Count%20=%201,%20MaxCookieSize%20=%201024,%20PerDomainCapacity%20=%205%20%7D,%20blobs:%20%5B...

[ViktorHofer]

Yes I'm investigating and was following up on that here: dotnet/corefx#21734

If I don't find the root cause by today I will just disable this one blob sanity check for CookieContainer. I still think it's somehow related to reflection...

[ViktorHofer]

@tarekgh we get different results when serializing a CookieContainer object which has a CompareInfo object in it on OSX. The serialized blob doesn't contain the m_name field value: https://www.diffchecker.com/wIO3OQ2U

This is the serialized object: new System.Net.CookieContainer(10, 5, 1024);

Any idea why this is happening? As you can read above this isn't happening on Darwin 16.6 but on Darwin 16.4.

[KristinXie1]

Failed again: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Filc~2F/build/20170705.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20AdjustmentRule%20%7B%20DateEnd%20=%201955-12-31T00:00:00.0000000,%20DateStart%20=%201900-01-01T00:00:00.0000000,%20Da...

[tarekgh]

we get different results when serializing a CookieContainer object which has a CompareInfo object in it on OSX. The serialized blob doesn't contain the m_name field value:

are we running on exactly same environment? I mean I am seeing in the original serialization data en-US is used. Are we using exactly the same on OSX? could be the used culture is the Invariant which has empty string name?

[ViktorHofer]

Yes, it's the same test which passed on Windows but failed on OSX (Darwin 16.4). I'm not 100% sure which Culture is used for the CookieContainer. Will check later.

[danmoseley]

I saw this cookiecontainer failure on a Unix run yesterday..

[ViktorHofer]

Could be. But what I'm asking Tarek here is about the missing field name value. If you saw the test falling it could be something else in the blob.

[danmoseley]

I can't find the Unix failure now -- it was in a PR. I see SortedList failing in 2.0 now (I guess this run had the change to skip CookieContainer):

https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170705.02/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20%5B%5D,%20blobs:%20%5B%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAEAQAAAB1TeXN0ZW0uQ29sbGVjdG%5C%22...,%20%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAEAQ...

The stored blob for type System.Collections.SortedList is outdated and needs to be updated.\nStored blob: AAEAAAD/////AQAAAAAAAAAEAQAAAB1TeXN0Z...

[danmoseley]

Aha, here's the CookieContainer failure showing up in Windows 10 and in Fedora 26 (before you disabled the check, obviusly)

https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170705.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20CookieContainer%20%7B%20Capacity%20=%2010,%20Count%20=%201,%20MaxCookieSize%20=%201024,%20PerDomainCapacity%20=%205%20%7D,%20blobs:%20%5B...

[joperezr]

Assinging to @ViktorHofer since he was looking into this.

[v-haren]

failed again: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Fcli~2F/build/20170719.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20CookieContainer%20%7B%20Capacity%20=%2010,%20Count%20=%201,%20MaxCookieSize%20=%201024,%20PerDomainCapacity%20=%205%20%7D,%20blobs:%20%5B...

[ViktorHofer]

Yes, but I'm currently at an internal event. I can look into it earliest by Sunday.

[ViktorHofer]

@safern The log you posted is a different issue (AdjustmentRule type, not CookieContainer type).

[safern]

@safern The log you posted is a different issue (AdjustmentRule type, not CookieContainer type).

You're right. Deleting my comment.

[KristinXie1]

This issue is repro on UWP F5 Tests (UWP6.0), detail: https://mc.dot.net/#/product/netcore/uwp60/source/official~2Fcorefx~2Frelease~2Fuwp6.0~2F/type/test~2Ffunctional~2Fuwp~2F/build/20170722.02/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20UTC,%20blobs:%20%5B%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMAgAAAE5TeXN0ZW0uQ29yZSwgVm%5C%22...,%20%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMA...

[danmoseley]

The failure now is The stored blob for type System.TimeZoneInfo is outdated and needs to be updated

https://mc.dot.net/#/product/netcore/uwp60/source/official~2Fcorefx~2Frelease~2Fuwp6.0~2F/type/test~2Ffunctional~2Fuwp~2F/build/20170725.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20UTC,%20blobs:%20%5B%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMAgAAAE5TeXN0ZW0uQ29yZSwgVm%5C%22...,%20%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMA...

@tarekgh 's workaround in ae66307 looks like it was for TimeZoneInfo.AdjustmentRule only.

@ViktorHofer @tarekgh do we need to add TimeZoneInfo itself as a workaround?
Also do we plan to converge the implementations (the "real fix") in UWP60 or only in master?

[tarekgh]

@ViktorHofer may need to investigate why we are getting this serialization failure and possibly fix the test. We decided the code unification effort will not be done for UWP6.0

[danmoseley]

Moved out of UWP milestone.

[ViktorHofer]

Code unification of TimeZoneInfo is a non goal for UWP6.0 but for .NET Core 2.1.

The test wasn't failing before and I am not sure why it's failing now. So the required work here is to

• fix the test in uwp6.0
• Finish converging TimeZoneInfo code and update blobs in master

I'm on it.

[ViktorHofer]

Found the failure. The Culture=neutral casing is different (which I fixed a while ago).

Stored blob:
\0\u0001\0\0\0????\u0001\0\0\0\0\0\0\0\f\u0002\0\0\0NSystem.Core, Version=3.5.0.0, Culture=neutral

Runtime generated blob:
\0\u0001\0\0\0????\u0001\0\0\0\0\0\0\0\f\u0002\0\0\0NSystem.Core, Version=3.5.0.0, Culture=Neutral

I'm not exactly sure which codebase the UWP F5 tests are picking up as both coreclr and corert TimeZoneInfo versions have the lower cased Culture=neutral which is correct.

[ViktorHofer]

I thought the F5 tests would use the coreclr release/uwp6.0 branch: https://github.com/dotnet/coreclr/blob/release/uwp6.0/src/mscorlib/src/System/TimeZoneInfo.cs#L32

[ViktorHofer]

I can't repro that locally. It seems to be somehow related to a CI package mismatch. Closing for now.

[KristinXie1]

This issue is still repro on Core Tests (Master), detail: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Fcli~2F/build/20170727.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20CookieContainer%20%7B%20Capacity%20=%2010,%20Count%20=%201,%20MaxCookieSize%20=%201024,%20PerDomainCapacity%20=%205%20%7D,%20blobs:%20%5B...

[danmoseley]

@ViktorHofer decoded, basically same as I noted above earlier,

<<m_SortVersionculture ? System.Globalization.SortVersion???en-US
>>m_SortVersionculture ? System.Globalization.SortVersion

On OSX both versions, the en-US is expected but it is not present causing the test to fail.

[ViktorHofer]

It's a long standing issue now... Do we have OSX 10.12/10.13 machines I can connect to? I don't have one myself. I will probably also need help from @tarekgh.

[tarekgh]

@ViktorHofer let me know how I can help you here

[danmoseley]

I assume @tarekgh has a machine as he has some other 10.13 issues to take care of. If not Priya does.

[tarekgh]

I don't have 10.13 machine but we had a remote which can be used. I'll communicate the details to @ViktorHofer

[KristinXie1]

This issue is still repro, detail: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Fcli~2F/build/20170818.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20abc,%20blobs:%20%5B%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMAgAAAE5TeXN0ZW0uRGF0YSwgVm%5C%22...,%20%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMA...

[tarekgh]

@ViktorHofer do you know if this run has your fix?

[ViktorHofer]

The PR was merged 22 hours ago so I expect the fix to be in it. But setting the culture helped because CookieContainer isn't failing anymore! Now DataTable is failing, I just discovered that the serialized xml contains a line which stores the culture:

<xs:element name="DataSet123" msdata:IsDataSet="true" msdata:Locale="en-US">

I need to update the stored blobs. I don't know why that only fails on OSX.

[tarekgh]

I need to update the stored blobs. I don't know why that only fails on OSX.

most likely is the OSX machine doesn't have en-US as the user language and instead, we use invariant.

[ViktorHofer]

So should I instead use Invariant for all the tests?

[tarekgh]

You need to ensure having the same culture used in the serialized stream and when the test runs. If you can enforce specific culture in the serialized stream and have the same culture used in test validation that will guarantee same results. choosing Invariant is good.

[ViktorHofer]

Still failing: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Fcli~2F/build/20170819.01/workItem/System.Runtime.Serialization.Formatters.Tests/analysis/xunit/System.Runtime.Serialization.Formatters.Tests.BinaryFormatterTests~2FValidateAgainstBlobs(obj:%20System.Data.DataSet,%20blobs:%20%5B%5C%22AAEAAAD~2F~2F~2F~2F~2FAQAAAAAAAAAMAgAAAE5TeXN0ZW0uRGF0YSwgVm%5C%22...,%20%5C%22AAEAAAD~2F~2F...

@tarekgh why is OSX always behaving differently? Now with InvariantCulture the difference is the following:

Stored DataSet blob line:
<xs:element name="DataSet123" msdata:IsDataSet="true" msdata:Locale="">

Runtime generated DataSet blob line:
<xs:element name="DataSet123" msdata:IsDataSet="true" msdata:UseCurrentLocale="true">

[tarekgh]

Looking at the code

https://github.com/dotnet/corefx/blob/81506698c8355a23a1c5269b5fdeebd6fa94d687/src/System.Data.Common/src/System/Data/xmlsaver.cs#L1044

the check

                    !dt.Locale.Equals(CultureInfo.CurrentCulture)

I believe the OSX run has current culture is invariant which will cause setting

        _dsElement.SetAttribute(Keywords.MSD_USECURRENTLOCALE, Keywords.MSDNS, Keywords.TRUE);

Your test has to handle this as the type is producing different serialization data depending on the current culture.

[danmoseley]

BTW @ViktorHofer maybe improve ValidateAgainstBlobs to log the decoded blobs .. would be easier since I don't thikn we've seen the last of ValidateAgainstBlobs...