bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Mark H Weaver]

Our 'ghc-case-insensitive' package contains two 'inputs' field
initializers.  If I'm not mistaken, the first one is being effectively
ignored, so 'ghc-hunit' is not actually an input.

It would be good to clean this up so that we can start raising errors
when duplicate field initializers are present.

      Mark

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Ludovic Courtès]

Hi Mark,

Mark H Weaver <mhw@netris.org> skribis:

> Our 'ghc-case-insensitive' package contains two 'inputs' field
> initializers.  If I'm not mistaken, the first one is being effectively
> ignored, so 'ghc-hunit' is not actually an input.
>
> It would be good to clean this up so that we can start raising errors
> when duplicate field initializers are present.

What about applying your patch that detects duplicate fields, rebuilding
all the .go files, and then resolving any such case by keeping only the
last occurrence of the field?

This is what happens currently anyway so it won’t change the derivations
produced for the faulty packages.

Ludo’.

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Mark H Weaver]

Hi Ludovic,

ludo@gnu.org (Ludovic Courtès) writes:

> Mark H Weaver <mhw@netris.org> skribis:
>
>> Our 'ghc-case-insensitive' package contains two 'inputs' field
>> initializers.  If I'm not mistaken, the first one is being effectively
>> ignored, so 'ghc-hunit' is not actually an input.
>>
>> It would be good to clean this up so that we can start raising errors
>> when duplicate field initializers are present.
>
> What about applying your patch that detects duplicate fields, rebuilding
> all the .go files, and then resolving any such case by keeping only the
> last occurrence of the field?
>
> This is what happens currently anyway so it won’t change the derivations
> produced for the faulty packages.

It's true that it wouldn't change anything to simply remove those
ignored duplicate field initializers.  However, I thought it would be
better to give people familiar with these packages an opportunity to
investigate.  Someone may have had a good reason for adding those
inputs, even if they are not strictly needed for a successful build.

Of course, at some point we should timeout on this.  I would advocate
commenting out the redundant duplicates instead of simply deleting them,
along with a FIXME comment asking someone to investigate.  We could also
look in the commit history to find out who added those redundant inputs,
and ask them directly.

If you're impatient to get the duplicate field detection patch committed
soon, I could implement these "timeout" measures in the next couple of
days.

What do you think?

      Mark

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Ludovic Courtès]

Hi Mark,

Mark H Weaver <mhw@netris.org> skribis:

> If you're impatient to get the duplicate field detection patch committed
> soon, I could implement these "timeout" measures in the next couple of
> days.
>
> What do you think?

A timeout of a few days would be perfect.  :-)

Thank you!

Ludo’.

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Ludovic Courtès]

Hello Mark,

Mark H Weaver <mhw@netris.org> skribis:

> It's true that it wouldn't change anything to simply remove those
> ignored duplicate field initializers.  However, I thought it would be
> better to give people familiar with these packages an opportunity to
> investigate.  Someone may have had a good reason for adding those
> inputs, even if they are not strictly needed for a successful build.
>
> Of course, at some point we should timeout on this.  I would advocate
> commenting out the redundant duplicates instead of simply deleting them,
> along with a FIXME comment asking someone to investigate.  We could also
> look in the commit history to find out who added those redundant inputs,
> and ask them directly.
>
> If you're impatient to get the duplicate field detection patch committed
> soon, I could implement these "timeout" measures in the next couple of
> days.

I think we should go ahead now and mechanically fix packages with
duplicate fields, and apply the duplicate detection patch.

Would you like to do that?

Thanks,
Ludo’.

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Ludovic Courtès]

Ping!

I’d really like to apply the patch you posted in
<https://lists.gnu.org/archive/html/guix-devel/2018-04/msg00201.html>
along with the corresponding package fixes.

I can do that if you want.

Ludo’.

ludo@gnu.org (Ludovic Courtès) skribis:

> Hello Mark,
>
> Mark H Weaver <mhw@netris.org> skribis:
>
>> It's true that it wouldn't change anything to simply remove those
>> ignored duplicate field initializers.  However, I thought it would be
>> better to give people familiar with these packages an opportunity to
>> investigate.  Someone may have had a good reason for adding those
>> inputs, even if they are not strictly needed for a successful build.
>>
>> Of course, at some point we should timeout on this.  I would advocate
>> commenting out the redundant duplicates instead of simply deleting them,
>> along with a FIXME comment asking someone to investigate.  We could also
>> look in the commit history to find out who added those redundant inputs,
>> and ask them directly.
>>
>> If you're impatient to get the duplicate field detection patch committed
>> soon, I could implement these "timeout" measures in the next couple of
>> days.
>
> I think we should go ahead now and mechanically fix packages with
> duplicate fields, and apply the duplicate detection patch.
>
> Would you like to do that?
>
> Thanks,
> Ludo’.

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Mark H Weaver]

Hi Ludovic,

ludo@gnu.org (Ludovic Courtès) writes:

> Ping!
>
> I’d really like to apply the patch you posted in
> <https://lists.gnu.org/archive/html/guix-devel/2018-04/msg00201.html>
> along with the corresponding package fixes.

Yes, sorry for the long delay on this.  I've been nervous about
<https://bugs.gnu.org/29774>, which I encountered after applying my
patch to detect duplicate fields, and which has not yet been explained
as far as I know.  The problem seems to be sensitive to seemingly
unrelated changes in the source tree.  Since I run Guix from my own
private branch, I'm not confident what will happen if I apply this patch
to master.

> I can do that if you want.

Please do.

     Thanks!
       Mark

bug#31319: ghc-case-insensitive: Duplicate 'inputs' field.

[Ludovic Courtès]

Mark H Weaver <mhw@netris.org> skribis:

> Our 'ghc-case-insensitive' package contains two 'inputs' field
> initializers.  If I'm not mistaken, the first one is being effectively
> ignored, so 'ghc-hunit' is not actually an input.

Fixed in b2e46ac5318dd790530659a459bd1e714287ff37.

Ludo’.