From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 16 Jul 2014 00:49:30 GMT
Subject: Re: [PATCH v8 1/4] usage: make error functions a stack
Message-ID: <CAPc5daUhmeucDOakXpR_ViS0cfFDEyXUD_xoeP6qeCg33dW+1A@mail.gmail.com>
In-Reply-To: <1405466783.2577.15.camel@jekeller-desk1.amr.corp.intel.com>

I actually am not a big fan of "stack" for a thing like this, to be honest.
Wouldn't it be sufficient for the callers who want specific behaviour from
its callees to

 - save away the current error/warning routines;
 - set error/warning routines to its own custom versions;
 - call the callees;
 - set error/warning routines back to their original; and
 - return from it

at least in the code paths under discussion?


On Tue, Jul 15, 2014 at 4:26 PM, Keller, Jacob E
<jacob.e.keller@intel.com> wrote:
> On Tue, 2014-07-15 at 15:47 -0700, Junio C Hamano wrote:
>> Jacob Keller <jacob.e.keller@intel.com> writes:
>>
>> >  extern void set_error_routine(void (*routine)(const char *err, va_list params));
>> > +extern void pop_error_routine(void);
>>
>> pop that undoes set smells somewhat weird.  Perhaps we should rename
>> set to push?  That would allow us catch possible topics that add new
>> calls to set_error_routine() as well by forcing the system not to
>> link when they are merged without necessary fixes.
>>
>
> Also curious what your thoughts on making every set_*_routine to be a
> stack? For now I was only planning on error but it maybe makes sense to
> change them all?
>
> Thanks,
> Jake
>