Re: Jakarta Mail TCK - Additional Thoughts? (was: TomEE 9.x - from javax to jakarta namespace)

[Zowalla, Richard]

Hi Romain,

thanks for the pointer - it sounds somehow familiar to what we
observed. Need to check though :)

Gruß
Richard

Am Donnerstag, dem 02.06.2022 um 09:17 +0200 schrieb Romain Manni-
Bucau:
> Hi,
> 
> Did you try handling LITERAL+ capability (1)? I don't think we do as
> of
> today.
> 
> (1)
> https://datatracker.ietf.org/doc/html/rfc7888#:~:text=LITERAL%2B%20allows%20the%20alternate%20form%20of%20literals%20(called%20%22non%2D,are%204096%20bytes%20or%20less
> .
> Romain Manni-Bucau
> @rmannibucau  |  Blog
>  | Old Blog
>  | Github <
> https://github.com/rmannibucau> |
> LinkedIn  | Book
> <
> https://www.packtpub.com/application-development/java-ee-8-high-performance
> >
> 
> 
> Le mar. 31 mai 2022 à 09:54, Zowalla, Richard <
> richard.zowa...@hs-heilbronn.de> a écrit :
> 
> > Hi,
> > 
> > short update on this:
> > 
> > Collaborated with JL and exchanged some ideas via Slack.
> > 
> > We now tested James + Greenmail as mail servers to rule out any
> > hard-
> > coded TCK assumption regarding James. Both fail with the same
> > exception
> > / issue on the same TCK mail:
> > https://github.com/eclipse-ee4j/mail-tck/blob/2.0.0/tests/mailboxes/test1/9
> > 
> > The difference between the RI and our impl is basically the literal
> > header:
> > 
> > a5 APPEND test1 () "8-Dec-1996 15:30:12 +0100" {150432}
> > a5 BAD APPEND failed. Illegal arguments.
> > 
> > vs (RI):
> > 
> > A6 APPEND test1 () "08-Dec-1996 15:30:12 +0100" {153113+}
> > A6 OK [APPENDUID 466034631 1] APPEND completed.
> >   Copied 1 messages
> > 
> > I pushed a configured Jakarta Mail TCK 2.0.1 setup with updated
> > instructions into this repository: https://github.com/rzo1/mail-tck
> > 
> > In addition, I am CC'ing the geronimo list, in case some people
> > there
> > have additional ideas. Otherwise, we will need to take a dive into
> > the
> > imap spec / server-side impl to get any clues :)
> > 
> > Gruß
> > Richard
> > 
> > 
> > Am Dienstag, dem 24.05.2022 um 19:46 + schrieb Zowalla,
> > Richard:
> > > Hi,
> > > 
> > > I spend some more time on the mail tck and got some additional
> > > insights:
> > > 
> > > There is one specific mail from the TCK mailbox (test1, mail no.
> > > 9),
> > > which breaks the current Geronimo mail impl. This happens, if you
> > > try
> > > to bootstrap / setup the test mailbox before running the TCK
> > > according
> > > ti their documentation. The same procedere just works, if the
> > > reference
> > > impl is used.
> > > 
> > > The failing tests in the mail tck report similar issues regarding
> > > failed IMAP commands. Therefore, I assume, that the underlying
> > > issue
> > > is
> > > similar, i.e. if we solve that, we likely fix some of the TCK
> > > tests
> > > too.
> > > 
> > > I added some instructions to
> > > https://issues.apache.org/jira/browse/GERONIMO-6835 to reproduce
> > > the
> > > issue without actually running the TCK, so we might have the
> > > chance
> > > to
> > > debug it easily.
> > > 
> > > Basically:
> > > 
> > > - Checkout 
> > > https://github.com/rzo1/geronimo-javamail/tree/tck-issues
> > > - Follow the instructions in tck.adoc to start up a mail server
> > > (docker-compose + docker exec)
> > > - Run "fpopulate" with arguments "-s test1 -d
> > > imap://user01%40james.local:1234@localhost:1143 -D" from within
> > > your
> > > IDE
> > > - Observe the debug output on the console
> > > 
> > > 
> > > There is a difference between the message length between the RI
> > > and
> > > the
> > > Geronimo impl (as reported by the { } literal). This might be the
> > > cause
> > > (??), but I have no idea what is going on or why it is happening.
> > > 
> > > Maybe someone has an idea what is going on here? Or has a pointer
> > > where
> > > to look at? I might be "lost in the tck madness" for today :)
> > > 
> > > Gruß
> > > Richard
> > > 
> > > 
> > > 
> > > Am Dienstag, dem 24.05.2022 um 17:13 + schrieb Zowalla,
> > > Richard:
> > > > To give a more detailed view / update from the spec tck party
> > > > regarding
> > > > activation and mail:
> > > > 
> > > > (A) Geronimo Activation 2.0
> > > > 
> > > > After a first milestone (M1) and some additional fixes after
> > > > running
> > > > the activation TCK [1] and related signatures tests, we are now
> > > > passing
> > > > them.
> > > > 
> > > > JL prepared a release artifact (1.0.0), which is currently
> > > > under
> > > > vote.
> > > > 
> > > > During the tck work, we found some inconsistency / unspecified
> > > > behaviour of "normalizeMimeTypeParameter" of
> > > > ActivationDataFlavor.
> > > > While this method is tested in the TCK on the basis of the
> > > > reference
> > > > implementation neither the spec itself nor the javadoc are
> > > > really
> > > > clear
> > > > about the "right" return value. At the moment, we adjusted it
> > > > to
> > > > pass
> > > > the TCK test in 

Re: Jakarta Mail TCK - Additional Thoughts? (was: TomEE 9.x - from javax to jakarta namespace)

[Romain Manni-Bucau]

Hi,

Did you try handling LITERAL+ capability (1)? I don't think we do as of
today.

(1)
https://datatracker.ietf.org/doc/html/rfc7888#:~:text=LITERAL%2B%20allows%20the%20alternate%20form%20of%20literals%20(called%20%22non%2D,are%204096%20bytes%20or%20less
.
Romain Manni-Bucau
@rmannibucau  |  Blog
 | Old Blog
 | Github  |
LinkedIn  | Book



Le mar. 31 mai 2022 à 09:54, Zowalla, Richard <
richard.zowa...@hs-heilbronn.de> a écrit :

> Hi,
>
> short update on this:
>
> Collaborated with JL and exchanged some ideas via Slack.
>
> We now tested James + Greenmail as mail servers to rule out any hard-
> coded TCK assumption regarding James. Both fail with the same exception
> / issue on the same TCK mail:
> https://github.com/eclipse-ee4j/mail-tck/blob/2.0.0/tests/mailboxes/test1/9
>
> The difference between the RI and our impl is basically the literal
> header:
>
> a5 APPEND test1 () "8-Dec-1996 15:30:12 +0100" {150432}
> a5 BAD APPEND failed. Illegal arguments.
>
> vs (RI):
>
> A6 APPEND test1 () "08-Dec-1996 15:30:12 +0100" {153113+}
> A6 OK [APPENDUID 466034631 1] APPEND completed.
>   Copied 1 messages
>
> I pushed a configured Jakarta Mail TCK 2.0.1 setup with updated
> instructions into this repository: https://github.com/rzo1/mail-tck
>
> In addition, I am CC'ing the geronimo list, in case some people there
> have additional ideas. Otherwise, we will need to take a dive into the
> imap spec / server-side impl to get any clues :)
>
> Gruß
> Richard
>
>
> Am Dienstag, dem 24.05.2022 um 19:46 + schrieb Zowalla, Richard:
> > Hi,
> >
> > I spend some more time on the mail tck and got some additional
> > insights:
> >
> > There is one specific mail from the TCK mailbox (test1, mail no. 9),
> > which breaks the current Geronimo mail impl. This happens, if you try
> > to bootstrap / setup the test mailbox before running the TCK
> > according
> > ti their documentation. The same procedere just works, if the
> > reference
> > impl is used.
> >
> > The failing tests in the mail tck report similar issues regarding
> > failed IMAP commands. Therefore, I assume, that the underlying issue
> > is
> > similar, i.e. if we solve that, we likely fix some of the TCK tests
> > too.
> >
> > I added some instructions to
> > https://issues.apache.org/jira/browse/GERONIMO-6835 to reproduce the
> > issue without actually running the TCK, so we might have the chance
> > to
> > debug it easily.
> >
> > Basically:
> >
> > - Checkout https://github.com/rzo1/geronimo-javamail/tree/tck-issues
> > - Follow the instructions in tck.adoc to start up a mail server
> > (docker-compose + docker exec)
> > - Run "fpopulate" with arguments "-s test1 -d
> > imap://user01%40james.local:1234@localhost:1143 -D" from within your
> > IDE
> > - Observe the debug output on the console
> >
> >
> > There is a difference between the message length between the RI and
> > the
> > Geronimo impl (as reported by the { } literal). This might be the
> > cause
> > (??), but I have no idea what is going on or why it is happening.
> >
> > Maybe someone has an idea what is going on here? Or has a pointer
> > where
> > to look at? I might be "lost in the tck madness" for today :)
> >
> > Gruß
> > Richard
> >
> >
> >
> > Am Dienstag, dem 24.05.2022 um 17:13 + schrieb Zowalla, Richard:
> > > To give a more detailed view / update from the spec tck party
> > > regarding
> > > activation and mail:
> > >
> > > (A) Geronimo Activation 2.0
> > >
> > > After a first milestone (M1) and some additional fixes after
> > > running
> > > the activation TCK [1] and related signatures tests, we are now
> > > passing
> > > them.
> > >
> > > JL prepared a release artifact (1.0.0), which is currently under
> > > vote.
> > >
> > > During the tck work, we found some inconsistency / unspecified
> > > behaviour of "normalizeMimeTypeParameter" of ActivationDataFlavor.
> > > While this method is tested in the TCK on the basis of the
> > > reference
> > > implementation neither the spec itself nor the javadoc are really
> > > clear
> > > about the "right" return value. At the moment, we adjusted it to
> > > pass
> > > the TCK test in question.
> > >
> > > There is an ongoing discussion at dev@geronimo if this is a desired
> > > behaviour or if a system property should be introduced in order to
> > > reduce the possibility of breaking some users.
> > >
> > > (B) Geronimo Mail 2.0 / 2.1
> > >
> > > The current mail impl has some TCK failures. It seems, that we need
> > > to
> > > do some additional work to get it compliant with the standalone
> > > mail
> > > tck [3].
> > >
> > > The signature tests are failing for Java 11 but are fine with Java
> > > 8
> > > [4] due to some usage of Object#finalize() and missing annotations
> > > (only 

Re: Jakarta Mail TCK - Additional Thoughts? (was: TomEE 9.x - from javax to jakarta namespace)

[Zowalla, Richard]

Hi,

short update on this:

Collaborated with JL and exchanged some ideas via Slack.

We now tested James + Greenmail as mail servers to rule out any hard-
coded TCK assumption regarding James. Both fail with the same exception
/ issue on the same TCK mail: 
https://github.com/eclipse-ee4j/mail-tck/blob/2.0.0/tests/mailboxes/test1/9

The difference between the RI and our impl is basically the literal
header:

a5 APPEND test1 () "8-Dec-1996 15:30:12 +0100" {150432}
a5 BAD APPEND failed. Illegal arguments.

vs (RI):

A6 APPEND test1 () "08-Dec-1996 15:30:12 +0100" {153113+}
A6 OK [APPENDUID 466034631 1] APPEND completed.
  Copied 1 messages

I pushed a configured Jakarta Mail TCK 2.0.1 setup with updated
instructions into this repository: https://github.com/rzo1/mail-tck 

In addition, I am CC'ing the geronimo list, in case some people there
have additional ideas. Otherwise, we will need to take a dive into the
imap spec / server-side impl to get any clues :)

Gruß
Richard


Am Dienstag, dem 24.05.2022 um 19:46 + schrieb Zowalla, Richard:
> Hi,
> 
> I spend some more time on the mail tck and got some additional
> insights:
> 
> There is one specific mail from the TCK mailbox (test1, mail no. 9),
> which breaks the current Geronimo mail impl. This happens, if you try
> to bootstrap / setup the test mailbox before running the TCK
> according
> ti their documentation. The same procedere just works, if the
> reference
> impl is used.
> 
> The failing tests in the mail tck report similar issues regarding
> failed IMAP commands. Therefore, I assume, that the underlying issue
> is
> similar, i.e. if we solve that, we likely fix some of the TCK tests
> too.
> 
> I added some instructions to 
> https://issues.apache.org/jira/browse/GERONIMO-6835 to reproduce the
> issue without actually running the TCK, so we might have the chance
> to
> debug it easily. 
> 
> Basically:
> 
> - Checkout https://github.com/rzo1/geronimo-javamail/tree/tck-issues
> - Follow the instructions in tck.adoc to start up a mail server
> (docker-compose + docker exec)
> - Run "fpopulate" with arguments "-s test1 -d
> imap://user01%40james.local:1234@localhost:1143 -D" from within your
> IDE
> - Observe the debug output on the console
> 
> 
> There is a difference between the message length between the RI and
> the
> Geronimo impl (as reported by the { } literal). This might be the
> cause
> (??), but I have no idea what is going on or why it is happening.
> 
> Maybe someone has an idea what is going on here? Or has a pointer
> where
> to look at? I might be "lost in the tck madness" for today :)
> 
> Gruß
> Richard
> 
> 
> 
> Am Dienstag, dem 24.05.2022 um 17:13 + schrieb Zowalla, Richard:
> > To give a more detailed view / update from the spec tck party
> > regarding
> > activation and mail:
> > 
> > (A) Geronimo Activation 2.0
> > 
> > After a first milestone (M1) and some additional fixes after
> > running
> > the activation TCK [1] and related signatures tests, we are now
> > passing
> > them. 
> > 
> > JL prepared a release artifact (1.0.0), which is currently under
> > vote.
> > 
> > During the tck work, we found some inconsistency / unspecified
> > behaviour of "normalizeMimeTypeParameter" of ActivationDataFlavor.
> > While this method is tested in the TCK on the basis of the
> > reference
> > implementation neither the spec itself nor the javadoc are really
> > clear
> > about the "right" return value. At the moment, we adjusted it to
> > pass
> > the TCK test in question.
> > 
> > There is an ongoing discussion at dev@geronimo if this is a desired
> > behaviour or if a system property should be introduced in order to
> > reduce the possibility of breaking some users.
> > 
> > (B) Geronimo Mail 2.0 / 2.1
> > 
> > The current mail impl has some TCK failures. It seems, that we need
> > to
> > do some additional work to get it compliant with the standalone
> > mail
> > tck [3].
> > 
> > The signature tests are failing for Java 11 but are fine with Java
> > 8
> > [4] due to some usage of Object#finalize() and missing annotations
> > (only available in Java 9+) in the Geronimo implementation. While
> > it
> > is
> > not that important for EE9, we need to keep it in mind for EE10.
> > 
> > We currently pass 166 out of 321 mail tck tests [5]. I guess, we
> > need
> > to give it some more love to get the numbers up and finally get it
> > to
> > pass the mail tck. The good thing is, that we already pass the
> > javamail
> > tests for TomEE [6].
> > 
> > Gruß
> > Richard
> > 
> > 
> > 
> > [1] https://jakarta.ee/specifications/activation/2.0/
> > [2] 
> > https://lists.apache.org/thread/h8twm4rmdxt67fx227nyywjp96b6cky1
> > [3] https://jakarta.ee/specifications/mail/2.0/
> > [4] https://issues.apache.org/jira/browse/GERONIMO-6834
> > [5] https://issues.apache.org/jira/browse/GERONIMO-6835
> > [6]  
> > https://tck.work/tomee/tests?build=1651841331620=com.sun.ts.tests.javamail
> > 
> > Am Dienstag, dem 24.05.2022 um 15:44 +0200 schrieb