stringify frames per as_string (#19) #20
Conversation
|
I have no idea why tests fail. They succeed locally. |
|
I think you used an older version of Perl::Tidy locally to tidy your code. You need to use the latest version from CPAN. |
|
Yes perltidy was the problem, but the real problem was that perlbrew was using the wrong perl. I must have said |
| @@ -3,6 +3,8 @@ package Devel::StackTrace::Frame; | |||
| use strict; | |||
| use warnings; | |||
|
|
|||
| use DDP; | |||
There was a problem hiding this comment.
Quick replies while I'm on holiday. I'll fix the problems next week when I get back.
I wasn't aware that I had committed generated files. Sorry!
Also apologies for forgetting to weed out (all) my debugging stuff.
| @@ -28,6 +30,11 @@ BEGIN { | |||
| } | |||
| } | |||
|
|
|||
| # XXX -- Arguments are garbled if giving overload \&as_string or 'as_string' directly. | |||
There was a problem hiding this comment.
It means that the test for the stringification being equal to the return value of the as_string method being equal fails if the value of the '""' key of use overload is \&as_string or 'as_string': part of the string appears as a ? in the stringification. I guess it's because some of the generated subs don't exist yet when use overload is executed. The form wrapping the method call in an anonymous sub does not have this problem and the test succeeds.
See #19.
I had to give overload an anonymous sub which calls
as_stringon the object rather than giving it\&as_stringor'as_string'or arguments wouldn't be shown properly in the stringification. Probably because of generated accessors which don't exist yet?I added a very basic test checking that the stringification is identical to the return value of
$frame->as_string. It caught the above problem so apparently not too simple!