Pure print callback adder function.

Pure print callback adder function.



Pure print callbacks

[Carl, 2010-09-06 10:35:36]

LGTM.  

The else clause is one space too far to the right,

I couldn't do sn inline comment because the side-by-side diff is missing.

Carl

[mikesol_ufl.edu, 2010-09-06 12:48:56]

Thanks Carl,
    I am reticent to go any farther with this patch after Joe's email
concerning what pure print functions actually need to do.  There is no way
to verify that these added functions behave correctly, which means that we
put the user in a position where they'll see no warning or error messages
but, like me, will be waiting with their fingers crossed for upwards of 13
hours before realizing that lilypond had descended into an infinite loop
because of various callbacks that their home-cooked pure print function
triggered without their knowing it.  Of course, we could just write
documentation to alert users of this, but there is always the chance that
(1) they don't read it; (2) even after having read it, they aren't familiar
enough with lilypond's internal workings to know if what they're doing is,
in fact, triggering one of the callbacks that pure print cannot touch.
    None of this is necessarily an impediment to moving forward if you feel
that the user should just be careful, but it does run the risk of leaving an
unchecked way in lilypond to cause a huge time drain.

~Mike


On 9/6/10 6:35 AM, "Carl.D.Sorensen@gmail.com" <Carl.D.Sorensen@gmail.com>
wrote:

> LGTM.
> 
> The else clause is one space too far to the right,
> 
> I couldn't do sn inline comment because the side-by-side diff is
> missing.
> 
> Carl
> 
> http://codereview.appspot.com/2020041/
> 

[c_sorensen, 2010-09-06 17:00:54]

On 9/6/10 6:53 AM, "Mike Solomon" <mikesol@ufl.edu> wrote:

> Thanks Carl,
>     I am reticent to go any farther with this patch after Joe's email
> concerning what pure print functions actually need to do.  There is no way
> to verify that these added functions behave correctly, which means that we
> put the user in a position where they'll see no warning or error messages
> but, like me, will be waiting with their fingers crossed for upwards of 13
> hours before realizing that lilypond had descended into an infinite loop
> because of various callbacks that their home-cooked pure print function
> triggered without their knowing it.  Of course, we could just write
> documentation to alert users of this, but there is always the chance that
> (1) they don't read it; (2) even after having read it, they aren't familiar
> enough with lilypond's internal workings to know if what they're doing is,
> in fact, triggering one of the callbacks that pure print cannot touch.
>     None of this is necessarily an impediment to moving forward if you feel
> that the user should just be careful, but it does run the risk of leaving an
> unchecked way in lilypond to cause a huge time drain.


Ahh -- now I understand the related emails.  I agree with you.  Until we can
get a really good definition of what is  allowed in pure print callbacks, we
*shouldn't* make the list public.

OK, I'll add a comment to the tracker and move the priority to postponed.

Thanks,

Carl

> 
> ~Mike
> 
> 
> On 9/6/10 6:35 AM, "Carl.D.Sorensen@gmail.com" <Carl.D.Sorensen@gmail.com>
> wrote:
> 
>> LGTM.
>> 
>> The else clause is one space too far to the right,
>> 
>> I couldn't do sn inline comment because the side-by-side diff is
>> missing.
>> 
>> Carl
>> 
>> http://codereview.appspot.com/2020041/
>> 
> 
>