Re: Updates to fret-diagrams

[Carl D. Sorensen]

On 1/2/09 9:16 PM, "Joe Neeman" <address@hidden> wrote:

> On Saturday 03 January 2009 02:30:36 pm Carl D. Sorensen wrote:
>> Thanks for the review, Joe.
>> 
>> On 1/2/09 4:17 PM, "address@hidden" <address@hidden> wrote:
>>> Reviewers: Carl.D.Sorensen,
>>> 
>>> 
>>> http://codereview.appspot.com/11857/diff/1/2
>>> File input/regression/fret-diagrams.ly (right):
>>> 
>>> http://codereview.appspot.com/11857/diff/1/2#newcode1
>>> Line 1: \version "2.12.0"
>>> This regtest is getting quite large. Is there a logical way to split it
>>> up (eg. fret-diagrams-landscape, fret-diagrams-string-count, etc)?
>> 
>> The regtest can easily be split up.  Is there a reason to do so?  I would
>> think that any time a change was made to the fret diagram code, the whole
>> regtest would need to be run anyway.
>> 
>> I'm not trying to be argumentative.  I had the same impression, that the
>> regtest was too long, but I needed all of those tests to make sure that
>> everything worked properly.
> 
> If a regtest breaks, it's much easier to spot the problem (particularly if the
> person trying to spot it isn't familiar with fret diagrams) if the test is
> small. That way, when someone breaks string-count, say, then "make check" will
> complain about input/regression/fret-diagrams-string-count.ly and it is
> perfectly obvious which bits of the fret diagram code were broken and which
> weren't.
> 

OK, so should I create a separate regtest for each item in
fret-diagram-details?  I selected the diagrams to make sure that all of them
were covered, but they are all mixed in together.  I could imagine having
something on the order of 15 individual regtests for the fret diagrams.
Each one would then be very tightly focused.

Would that be preferable?  Or should I try to find some middle ground?  Is
there a problem if I have too many regtests?

Thanks,

Carl