T050-new and T060-count fail if gdb prints a warning

T050-new and T060-count fail if gdb prints a warning

[Dan Čermák]

[-- Attachment #1: Type: text/plain, Size: 1504 bytes --]

Hi list,

I have encountered a test failure caused by the T050-new and T060-count
tests in the latest openSUSE-Tumbleweed rebuild with Python 3.8 instead
of Python 3.7. Both tests die with:

--8<---------------cut here---------------start------------->8---
T050-new: Testing "notmuch new" in several variations
 FAIL   Handle files vanishing between scandir and add_file
       --- T050-new.36.EXPECTED        2020-03-10 12:49:34.388407058 +0000
       +++ T050-new.36.OUTPUT  2020-03-10 12:49:34.388407058 +0000
       @@ -1,3 +1,7 @@
       +/usr/share/gdb/python/gdb/command/prompt.py:48: SyntaxWarning: "is not" with a literal. Did you mean "!="?
       +  if self.value is not '':
       +/usr/share/gdb/python/gdb/command/prompt.py:60: SyntaxWarning: "is not" with a literal. Did you mean "!="?
       +  if self.value is not '':
        Unexpected error with file /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish
        add_file: Something went wrong trying to read or write a file
        Error opening /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish: No such file or directory
--8<---------------cut here---------------end--------------->8---

which I think is caused by gdb 8 using some odd Python syntax and is
completely unrelated to notmuch.

The solution is of course to fix this in gdb (e.g. by updating to gdb
9), but the more general question is: can notmuch's testsuite do
something to avoid issues like this altogether?


Cheers,

Dan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Re: T050-new and T060-count fail if gdb prints a warning

[David Bremner]

Dan Čermák <dan.cermak@posteo.net> writes:
>
> The solution is of course to fix this in gdb (e.g. by updating to gdb
> 9), but the more general question is: can notmuch's testsuite do
> something to avoid issues like this altogether?

I replaced some of the uses of gdb with LD_PRELOAD based shims. I'm not
sure how much work that would be to do here, but maybe someone(TM) would
like to try it? There is some examples in T070-insert.sh.

d

Re: T050-new and T060-count fail if gdb prints a warning

[Tomi Ollila]

On Tue, Mar 10 2020, Dan Čermák wrote:

> Hi list,
>
> I have encountered a test failure caused by the T050-new and T060-count
> tests in the latest openSUSE-Tumbleweed rebuild with Python 3.8 instead
> of Python 3.7. Both tests die with:
>
> --8<---------------cut here---------------start------------->8---
> T050-new: Testing "notmuch new" in several variations
>  FAIL   Handle files vanishing between scandir and add_file
>        --- T050-new.36.EXPECTED        2020-03-10 12:49:34.388407058 +0000
>        +++ T050-new.36.OUTPUT  2020-03-10 12:49:34.388407058 +0000
>        @@ -1,3 +1,7 @@
>        +/usr/share/gdb/python/gdb/command/prompt.py:48: SyntaxWarning: "is not" with a literal. Did you mean "!="?
>        +  if self.value is not '':
>        +/usr/share/gdb/python/gdb/command/prompt.py:60: SyntaxWarning: "is not" with a literal. Did you mean "!="?
>        +  if self.value is not '':
>         Unexpected error with file /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish
>         add_file: Something went wrong trying to read or write a file
>         Error opening /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish: No such file or directory
> --8<---------------cut here---------------end--------------->8---
>
> which I think is caused by gdb 8 using some odd Python syntax and is
> completely unrelated to notmuch.
>
> The solution is of course to fix this in gdb (e.g. by updating to gdb
> 9), but the more general question is: can notmuch's testsuite do
> something to avoid issues like this altogether?

setting PYTHONWARNINGS environment variable could probably do it.

IIRC the -count test is something that is harder to replace with
LD_PRELOAD shim...

(btw: thanks for this report -- i learned python3 -bb option
(been hit comparing strings and bytes with == ...) 

>
>
> Cheers,
>
> Dan


Tomi

Re: T050-new and T060-count fail if gdb prints a warning

[Dan Čermák]

[-- Attachment #1: Type: text/plain, Size: 2252 bytes --]

Hi Tomi,

Tomi Ollila <tomi.ollila@iki.fi> writes:

> On Tue, Mar 10 2020, Dan Čermák wrote:
>
>> Hi list,
>>
>> I have encountered a test failure caused by the T050-new and T060-count
>> tests in the latest openSUSE-Tumbleweed rebuild with Python 3.8 instead
>> of Python 3.7. Both tests die with:
>>
>> --8<---------------cut here---------------start------------->8---
>> T050-new: Testing "notmuch new" in several variations
>>  FAIL   Handle files vanishing between scandir and add_file
>>        --- T050-new.36.EXPECTED        2020-03-10 12:49:34.388407058 +0000
>>        +++ T050-new.36.OUTPUT  2020-03-10 12:49:34.388407058 +0000
>>        @@ -1,3 +1,7 @@
>>        +/usr/share/gdb/python/gdb/command/prompt.py:48: SyntaxWarning: "is not" with a literal. Did you mean "!="?
>>        +  if self.value is not '':
>>        +/usr/share/gdb/python/gdb/command/prompt.py:60: SyntaxWarning: "is not" with a literal. Did you mean "!="?
>>        +  if self.value is not '':
>>         Unexpected error with file /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish
>>         add_file: Something went wrong trying to read or write a file
>>         Error opening /home/abuild/rpmbuild/BUILD/notmuch-0.29.3/test/tmp.T050-new/mail/vanish: No such file or directory
>> --8<---------------cut here---------------end--------------->8---
>>
>> which I think is caused by gdb 8 using some odd Python syntax and is
>> completely unrelated to notmuch.
>>
>> The solution is of course to fix this in gdb (e.g. by updating to gdb
>> 9), but the more general question is: can notmuch's testsuite do
>> something to avoid issues like this altogether?
>
> setting PYTHONWARNINGS environment variable could probably do it.

TIL about this environment variable! Thanks a lot for the hint, that
fixed the build in openSUSE and is a far better temporary workaround
than skipping two tests.

>
> IIRC the -count test is something that is harder to replace with
> LD_PRELOAD shim...
>
> (btw: thanks for this report -- i learned python3 -bb option
> (been hit comparing strings and bytes with == ...)

Well, thanks for sharing this too, thereby I learned another thing today ;-)


Cheers,

Dan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]