Re: 13.47 regtests

[Carl Sorensen]

On 2/4/11 7:13 AM, "Phil Holmes" <address@hidden> wrote:

> "Phil Holmes" <address@hidden> wrote in message
> news:address@hidden
>> "Graham Percival" <address@hidden> wrote in message
>> news:address@hidden
>>> On Thu, Feb 03, 2011 at 06:34:55PM -0000, Phil Holmes wrote:
>>>> Has anyone checked them?
>>> 
>>> Probably not.
>>> 
>>> Also, there's currently 16 issues to verify.  Including some
>>> Critical issues.  All the bug squad needed to do was to copy&paste
>>> and example and check if it crashes or not.
>>> 
>>> I want to start building the second release candidate in 24 hours,
>>> but that would be silly if we have this many unconfirmed "fixed
>>> bugs".
>>> 
>>> - Graham
>> 
>> The initial report from my pixel-by-pixel comparison is that the treble
>> clef in .47 is slightly different from previously.  The lower downstroke
>> is thinner and the blob on the end of it is slightly different.  I've
>> attached a colour coded difference image, where the common pixels are
>> coloured red, and the differences are grey - look at the lower part of the
>> clef.
>> 
>> Is this intended?

Werner made changes in the clef to "make the code more readable".  It didn't
look to me like he was trying to make changes, but because the code was
refactored, it could have made small changes.

See commit 662f7fa0284ab9fae27942934c41354d64a5958e
 
>> 
>> Note that it also makes checking the pixel-by-pixel comparison slow going,
>> since it gives a false positive in almost every check.

Perhaps the pixel-by-pixel comparison isn't useful when a common font
characteristic changes.  The same thing would occur if a note head shape
changed (and that's not an impossibility).

You could probably work around it by checking out an old copy of the font
and running the .47 regtests with the old copy of the font.

HTH,

Carl