[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#1054494: RFS: lsp-mode/8.0.0-6 [ITA] [RC] -- Emacs client/library for the Language Server Protocol



Xiyue Deng <manphiz@gmail.com> writes:

> Hi Bo,
>
> Bo YU <tsu.yubo@gmail.com> writes:
>
>> Hi!
>>
>> On Thu, Oct 26, 2023 at 7:02 AM Xiyue Deng <manphiz@gmail.com> wrote:
>>
>> ...
>>>
>>> For the unlikely but possible cause that tests with a long name is a
>>> prefix of other tests that may trigger this issue, I have modified the
>>> test name for testing purposes.  Can you help get the latest upload on
>>> mentors and try again?  TIA.
>>>
>> I tried this and it seems the issue was raised with my sbuild build environment.
>> I still got this:
>> https://paste.debian.net/1296268/
>>
>> My sbuilderrc is here:
>> https://paste.debian.net/1296269/
>>
>> But if use pbuilder[0] to build your package, it is ok.
>> So I think your package which is no problem.
>>
>> BTW, I suspect the network accessing leads to the issue and I am annoy how to
>> disable network access during building for sbuild.
>>
>> BR,
>> Bo
>> [0]: https://wiki.ubuntu.com/PbuilderHowto
>
> Thanks for testing!  The reason I'm interested in reproducing this error
> is that in the report of the RC bug that this upload is trying to solve
> - https://bugs.debian.org/1052939 - the build log from Lucas has exactly
> the same error:
>
> ,----
> | ...
> | > Test ‘lsp-text-document-hover-request’ redefined
> | > 
> | > Error: error ("Test ‘lsp-text-document-hover-request’ redefined")
> | ...
> `----
>
> But I haven't been able to reproduce this until Arto and you sent your
> reports, and this being reproduced by two people makes this more
> interesting.  There must be something that may trigger this unlikely
> error.  Also I'm not sure whether the network accessing may have been
> the cause as sbuild needs to download the dependencies and without
> something like apt-cacher{,-ng} it does need network access for that to
> happen.
>
> I suspected that the parallel setting in $DEB_BUILD_OPTIONS may have
> affected it so I copied your sbuild settings and tried again but
> unfortunately it still succeeded for me.  For the unlikely event and for
> completeness, can you also try to turn that off in your sbuild config
> and retry just in case?  TIA.
>

Actually scratch my previous mail, as I found how to produce the issue.
In `lsp-clangd-test.el' it does `(require 'lsp-integration-test)', so if
`lsp-clangd-test.el' is loaded before `lsp-integration-test.el', it
seems the test symbols in the latter are loaded twice that triggers the
error regardless of whether there is an actual duplicated test name.  I
can confirm that in your build log that the clangd one was loaded first
which causes this error.  I assume Arto sees it due to the same cause.

I have added another change to override dh_elpa_test to ensure
`lsp-clangd-test' is loaded last and uploaded to mentors.  Please help
test again.

I'll probably also report this issue upstream to see how it should be
handled.

>>> ,----
>>> | $ dget -x https://mentors.debian.net/debian/pool/main/l/lsp-mode/lsp-mode_8.0.0-6.dsc
>>> | $ sbuild lsp-mode_8.0.0-6.dsc
>>> `----
>>>
>>> P.S. If you can provide the failed build log and ~/.sbuildrc it may
>>> still help to eliminate potential sbuild differences in our environment.
>>>
>>> >>
>>> >> BR,
>>> >> Bo
>>> >>>
>>> >>> --
>>> >>> Arto Jantunen
>>> >>>
>>>
>>> --
>>> Xiyue Deng

-- 
Xiyue Deng


Reply to: