You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@netbeans.apache.org by Michael Bien <mb...@gmail.com> on 2022/04/20 12:48:17 UTC
heads up: master is unstable
Hi devs,
if you take a look at the commit list:
https://github.com/apache/netbeans/commits/master
and check the reason travis is failing, you will notice its always the
same job and the same tests.
the failing test which the failing jobs have in common:
org.netbeans.modules.java.lsp.server.protocol.ServerTest
failed: testHover
errored: testMain
when did it start?
it appears to be caused by:
https://github.com/apache/netbeans/pull/3266
although it has a green check mark, travis did not run at all before
integration for some reason - at least its not listed there. 12/12
checks ran, it should have been 13.
I restarted it yesterday a few times but the queue was full, so I waited
a day before writing this mail.
best regards,
michael
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Michael Bien <mb...@gmail.com>.
On 21.04.22 09:39, Neil C Smith wrote:
> On Wed, 20 Apr 2022 at 20:45, Michael Bien <mb...@gmail.com> wrote:
>> Sandeep is working on the fix for master, ...
> That possibly wants to go via delivery now? If we want it in master
> quickly after that, we can do an extra sync across before the next RC.
i don't know, I only saw the PR. I hope it can be resolved soon though.
> On Wed, 20 Apr 2022 at 18:16, Michael Bien <mb...@gmail.com> wrote:
>> i am sure there used to be a reason why NB is build by three CI systems?
> But look at the glorious multi-coloured bikeshed! :-)
>
> I'm (still) in favour of moving down to 2 at least. The GitHub /
> Jenkins split makes some sense though.
so am I. I am always in favor of reducing complexity.
I will start moving jobs step by step from travis to gh then, once
everything is more quiet in a week or two.
If anyone is against this it would be good to speak up before that.
-michael
> Best wishes,
>
> Neil
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
> For additional commands, e-mail: dev-help@netbeans.apache.org
>
> For further information about the NetBeans mailing lists, visit:
> https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Neil C Smith <ne...@apache.org>.
On Wed, 20 Apr 2022 at 20:45, Michael Bien <mb...@gmail.com> wrote:
> Sandeep is working on the fix for master, ...
That possibly wants to go via delivery now? If we want it in master
quickly after that, we can do an extra sync across before the next RC.
On Wed, 20 Apr 2022 at 18:16, Michael Bien <mb...@gmail.com> wrote:
> i am sure there used to be a reason why NB is build by three CI systems?
But look at the glorious multi-coloured bikeshed! :-)
I'm (still) in favour of moving down to 2 at least. The GitHub /
Jenkins split makes some sense though.
Best wishes,
Neil
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Michael Bien <mb...@gmail.com>.
Yes, lets avoid merging for now and wait a bit.
Travis just started running some jobs again. Lets hope it recovers
quickly (queue is down to 50).
Sandeep is working on the fix for master, but the travis job of the PR
is still in queue atm:
https://github.com/apache/netbeans/pull/4014
this incident wasn't caused by unreliable tests, it was caused by the
fact that travis didn't even start for a PR - and it got integrated anyway:
https://github.com/apache/netbeans/pull/3266#issuecomment-1103966682
i am all for efforts of making tests more reliable though :)
best regards,
michael
On 20.04.22 20:44, antonio wrote:
> java.lsp.server tests are known to fail [1] since (at least) 2022-03-06.
>
> Basically we have a "while(true) {}" block in "ConnectionSpec" [2]
> that loops forever.
>
> We had a PR a while back, but it didn't attract much interest. A
> possible solution is to catch a SocketException and break the loop
> then as in [3].
>
> Also Google's GSON cannot deserialize different JSON objects coming
> from the LSP server through the wire, as the NetBeans POJOs don't have
> a proper constructor/are not created using a factory object. See [4]
> for an example. You can run "ant test" in java.lsp.server to verify this.
>
> I understand there's a lot of activity (and interest) in having
> java.lsp.server out quickly, but we may want to think twice how we got
> to this "master is unstable" situation, and how to proceed forward, so
> this does not happen again.
>
> As an immediate countermeasure I propose freezing all PR merges until
> master is stable.
>
> Cheers,
> Antonio
>
> [1]
> https://github.com/apache/netbeans/issues/3716
>
> [2]
> https://github.com/apache/netbeans/blob/385dbd3ca71f1643dde025deb18ed2e75a714005/java/java.lsp.server/src/org/netbeans/modules/java/lsp/server/ConnectionSpec.java#L112
>
>
> (Note this under the "src" folder, i.e., it's not a unit test file).
>
> [3]
> https://github.com/apache/netbeans/pull/3717/files#diff-99d2e9352dee99afecc8f129ce088efcbf7b3649f966a27e929dc7a8a2a507e6
>
>
> [4]
> https://github.com/apache/netbeans/pull/3717
>
> El 20/4/22 a las 14:48, Michael Bien escribió:
>>
>> the failing test which the failing jobs have in common:
>>
>> org.netbeans.modules.java.lsp.server.protocol.ServerTest
>> failed: testHover
>> errored: testMain
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
> For additional commands, e-mail: dev-help@netbeans.apache.org
>
> For further information about the NetBeans mailing lists, visit:
> https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by antonio <an...@vieiro.net>.
java.lsp.server tests are known to fail [1] since (at least) 2022-03-06.
Basically we have a "while(true) {}" block in "ConnectionSpec" [2] that
loops forever.
We had a PR a while back, but it didn't attract much interest. A
possible solution is to catch a SocketException and break the loop then
as in [3].
Also Google's GSON cannot deserialize different JSON objects coming from
the LSP server through the wire, as the NetBeans POJOs don't have a
proper constructor/are not created using a factory object. See [4] for
an example. You can run "ant test" in java.lsp.server to verify this.
I understand there's a lot of activity (and interest) in having
java.lsp.server out quickly, but we may want to think twice how we got
to this "master is unstable" situation, and how to proceed forward, so
this does not happen again.
As an immediate countermeasure I propose freezing all PR merges until
master is stable.
Cheers,
Antonio
[1]
https://github.com/apache/netbeans/issues/3716
[2]
https://github.com/apache/netbeans/blob/385dbd3ca71f1643dde025deb18ed2e75a714005/java/java.lsp.server/src/org/netbeans/modules/java/lsp/server/ConnectionSpec.java#L112
(Note this under the "src" folder, i.e., it's not a unit test file).
[3]
https://github.com/apache/netbeans/pull/3717/files#diff-99d2e9352dee99afecc8f129ce088efcbf7b3649f966a27e929dc7a8a2a507e6
[4]
https://github.com/apache/netbeans/pull/3717
El 20/4/22 a las 14:48, Michael Bien escribió:
>
> the failing test which the failing jobs have in common:
>
> org.netbeans.modules.java.lsp.server.protocol.ServerTest
> failed: testHover
> errored: testMain
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Michael Bien <mb...@gmail.com>.
On 20.04.22 17:15, Neil C Smith wrote:
> On Wed, 20 Apr 2022 at 13:50, Michael Bien <mb...@gmail.com> wrote:
>> although it has a green check mark, travis did not run at all before
>> integration for some reason - at least its not listed there. 12/12
>> checks ran, it should have been 13.
> Good catch! I've noticed that situation a few times over the last
> couple of weeks.
>
> .. next step, migrate everything else to GH??
you tell me, i am new here :)
i am sure there used to be a reason why NB is build by three CI systems?
but regarding travis:
I see 93 jobs in the queue and nothing is running (log in and look at
left panel, direct link doesn't really work). Yesterday it wasn't much
different. Unfortunately i can't find a timestamp anywhere so i can't
tell how old those jobs are.
Is something stuck? Should we cancel everything or would this make the
situation worse? Wait another day?
-michael
>
> Neil
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
> For additional commands, e-mail: dev-help@netbeans.apache.org
>
> For further information about the NetBeans mailing lists, visit:
> https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Neil C Smith <ne...@apache.org>.
On Wed, 20 Apr 2022 at 13:50, Michael Bien <mb...@gmail.com> wrote:
> although it has a green check mark, travis did not run at all before
> integration for some reason - at least its not listed there. 12/12
> checks ran, it should have been 13.
Good catch! I've noticed that situation a few times over the last
couple of weeks.
.. next step, migrate everything else to GH??
Neil
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Michael Bien <mb...@gmail.com>.
sorry please ignore what i just wrote
travis always applies the PR on top of latest master, while gh actions
will just test the given PR branch
so this won't help for travis.
On 21.04.22 00:21, Michael Bien wrote:
> travis caught up finally!
>
> If you have a PR active which was based on unstable master, you could
> rebase on top of 9d5defe761bfb6f967d87f4c03825cde6c1b1d45 which was
> the last stable commit on master. This should allow to give you green
> test results on PRs.
>
> But before integration, wait for a green master (#4014 attempts to fix
> it).
>
> regards,
> michael
>
>
> On 20.04.22 14:48, Michael Bien wrote:
>> Hi devs,
>>
>> if you take a look at the commit list:
>>
>> https://github.com/apache/netbeans/commits/master
>>
>> and check the reason travis is failing, you will notice its always
>> the same job and the same tests.
>>
>> the failing test which the failing jobs have in common:
>>
>> org.netbeans.modules.java.lsp.server.protocol.ServerTest
>> failed: testHover
>> errored: testMain
>>
>>
>> when did it start?
>>
>> it appears to be caused by:
>>
>> https://github.com/apache/netbeans/pull/3266
>>
>> although it has a green check mark, travis did not run at all before
>> integration for some reason - at least its not listed there. 12/12
>> checks ran, it should have been 13.
>>
>> I restarted it yesterday a few times but the queue was full, so I
>> waited a day before writing this mail.
>>
>> best regards,
>>
>> michael
>>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: heads up: master is unstable
Posted by Michael Bien <mb...@gmail.com>.
travis caught up finally!
If you have a PR active which was based on unstable master, you could
rebase on top of 9d5defe761bfb6f967d87f4c03825cde6c1b1d45 which was the
last stable commit on master. This should allow to give you green test
results on PRs.
But before integration, wait for a green master (#4014 attempts to fix it).
regards,
michael
On 20.04.22 14:48, Michael Bien wrote:
> Hi devs,
>
> if you take a look at the commit list:
>
> https://github.com/apache/netbeans/commits/master
>
> and check the reason travis is failing, you will notice its always the
> same job and the same tests.
>
> the failing test which the failing jobs have in common:
>
> org.netbeans.modules.java.lsp.server.protocol.ServerTest
> failed: testHover
> errored: testMain
>
>
> when did it start?
>
> it appears to be caused by:
>
> https://github.com/apache/netbeans/pull/3266
>
> although it has a green check mark, travis did not run at all before
> integration for some reason - at least its not listed there. 12/12
> checks ran, it should have been 13.
>
> I restarted it yesterday a few times but the queue was full, so I
> waited a day before writing this mail.
>
> best regards,
>
> michael
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@netbeans.apache.org
For additional commands, e-mail: dev-help@netbeans.apache.org
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists