bash evaluation change...

[Nathanael D. Noblet]

On 03/02/2011 04:23 PM, Nathanael D. Noblet wrote:
> On 03/02/2011 04:03 PM, Nathanael D. Noblet wrote:
>> Hello while trying to build a package of mine on F15 I ran into
>> configure issues. They stem from a change in how bash evaluates case
>> statements. My test script: http://fpaste.org/6zuv/
>>
>> F14:
>>
>> -Z!!!
>> = x
>> X1
>> X1.1<--- evaluates to No Match for F15
>> xyes2
>> xyes3
>>
>>
>> F15:
>>
>> -Z!!!
>> = x
>> X1
>> No match1.1<--- evaluates to X1.1 on F14
>> xyes2
>> xyes3
>>
>> Is this a regression that should be reported to bash? Or is this
>> expected and should be fixed in my package? You'll notice that the if
>> test methods continue to function in the same manner.
>
> Also my packages built fine for rawhide and F15 for quite awhile (last
> successful F15 build was Feb 8th). My guess when looking at bash builds
> in koji for F15 is that the transition from 4.1.9 to 4.2.0 caused the
> change. (it was build Feb 9th)....

One final post, the issue is when a variable *isn't* defined prior to 
use. This is the use case that has changed.