bug#29492: tests/guix-system.sh failure on unbound variable check

[Eric Bavier]

Pushed in 42d16037d857aac02add38513bfa58535c1ddcfe

Thanks,

Eric Bavier, Scientific Libraries, Cray Inc.

________________________________________
From: Ludovic Courtès <address@hidden>
Sent: Saturday, December 2, 2017 03:59
To: Eric Bavier
Cc: 'address@hidden'
Subject: Re: bug#29492: tests/guix-system.sh failure on unbound variable check

Eric Bavier <address@hidden> skribis:

>> -----Original Message-----
>> From: Ludovic Courtès [mailto:address@hidden
>> Sent: Thursday, November 30, 2017 4:04 AM
>> To: Eric Bavier
>> Cc: address@hidden
>> Subject: Re: bug#29492: tests/guix-system.sh failure on unbound variable
>> check
>>
>> Hi Eric,
>>
>> Eric Bavier <address@hidden> skribis:
>>
>> > Latest guix master (2cdf78df2d3d5d88c7e6908754233cf37cce1e61) fails
>> tests/guix-system.sh for me, on line 128.  This seems to be caused by the
>> fact that the error output contains a multi-character column number:
>> >
>> > ```
>> > /tmp/bavier/tmpfile:9:14: In procedure #<procedure 3aba2820 ()>:
>> > /tmp/bavier/tmpfile:9:14: GRUB-config: unbound variable
>> > hint: Did you forget a `use-modules' form?
>>
>> I suppose that’s with Guile 2.0, right?
>
> Right, 2.0.14.
>
>> So the patch would become:
>
> diff --git a/tests/guix-system.sh b/tests/guix-system.sh
> index 4bb866adf..eaa0c4332 100644
> --- a/tests/guix-system.sh
> +++ b/tests/guix-system.sh
> @@ -125,7 +125,8 @@ else
>       # See <http://bugs.gnu.org/26107>.
>       grep "$tmpfile:[49]:[0-9]: GRUB-config.*[Uu]nbound variable" 
> "$errorfile"
>      else
> -     grep "$tmpfile:9:[0-9]: GRUB-config.*[Uu]nbound variable" "$errorfile"
> +     # With Guile 2.0.14 the error is reported on line 14 (the last line).
> +     grep "$tmpfile:9:[0-9]\+: GRUB-config.*[Uu]nbound variable" "$errorfile"
>      fi
>  fi
>
> No, at *column* 14.  Which I believe is the desired result, right?  Character 
> 14 is the '(', the 'GRUB-config symbol itself starts at character 15.   But 
> now I wonder whether we should be using a regex for that anyhow.  Do we 
> expect the column number to change ever?

We don’t, but sometimes location info is not as precise as we’d like.

> I think it would be fine to fix the regex for Guile 2.0 only, but once the 
> bug affecting 2.2 is fixed, it'll need to be applied there too.  Maybe it 
> would make sense to fix both at the same time.

Yes, you’re right, it’s better to fix both.

Feel free to push.

Thank you!

Ludo’.