[Pharo-dev] [Bug] In ClassDescription>>#linesOfCode

Roberto Minelli roberto.minelli at usi.ch
Fri Nov 15 09:10:43 EST 2013


In fact *after* I wrote the email I actually thought about that.

IMHO we need to have a method which gave us only the LOC of the “side” we are interested in, i.e., there is no way to access only the LOC of the instance side and for me this is wrong. Something like

actualLinesOfCode

	^ self localMethods inject: 0 into: [:sum :each | sum + each linesOfCode]. 

and then

linesOfCode 
	"An approximate measure of lines of code.
	Includes comments, but excludes blank lines."
	| lines |
	lines := self actualLinesOfCode 
	^ self isMeta 
		ifTrue: [lines]
		ifFalse: [lines + self class linesOfCode]

So that one can invoke Object>>#linesOfCode and have the current behavior, or Object>>#actualLinesOfCode and have just the instance side lines of code, and Object class>>#actualLinesOfCode and get only the meta-lines of code.

Cheers,
R

On Nov 15, 2013, at 2:30 PM, Marcus Denker <marcus.denker at inria.fr> wrote:

> 
> On 15 Nov 2013, at 14:18, Marcus Denker <marcus.denker at inria.fr> wrote:
> 
>> 
>> On 15 Nov 2013, at 14:07, Roberto Minelli <roberto.minelli at usi.ch> wrote:
>> 
>>> Hi,
>>> 
>>> I looked at the implementation of #linesOfCode of ClassDescription. It is actually wrong, a correct version should be:
>>> 
>>> linesOfCode 
>>> 	"An approximate measure of lines of code. Includes comments, but excludes blank lines."
>>> 	| lines |
>>> 	
>>> 	lines := self localMethods inject: 0 into: [:sum :each | sum + each linesOfCode]. 
>>> 	^ self isMeta not
>>> 		ifTrue: [lines]
>>> 		ifFalse: [lines + self class linesOfCode]
>>> 
>> 
>> indeed!
>> 
> Hmm… not that sure after thinking about it.
> 
> On the meta side, we just take the linesOfCode of all methods. On the instance side, we take the linesOfCode of all method *and* the
> ones of the meta.
> 
> When you call 
> 
> Object linesOfCode
> 
> this means it takes
> 
> linesOfCode of Instance Side + linesOfCode of class side of Object
> 
> which seems ok.
> 
> 	Marcus
> 





More information about the Pharo-dev mailing list