[Scons-users] Is this a feature or a bug…

[William Deegan]

On 08/21/2012 08:40 AM, Chris BeHanna wrote:

> On Aug 21, 2012, at 10:04 , Gary Oberbrunner <garyo at oberbrunner.com> wrote:

>

>> On Tue, Aug 21, 2012 at 10:44 AM, Russel Winder <russel at winder.org.uk> wrote:

>>> Just asking the question prior to putting in a bug report, just in case

>>> I have missed something…

>>>

>>> Command(XXX, YYY, [AAA, BBB])

>>>

>>> i.e. the action sequence is a list, works fine.  However,

>>>

>>> Command(XXX, YYY, (AAA, BBB))

>>>

>>> is reported as an error, i.e. a tuple is not acceptable where a list is.

>>>

>>> To say this doesn't violate the Principle of Least Surprise would be to

>>> violate the Principle of Least Surprise.

>> I can't see any particular reason we shouldn't support tuples as well

>> as lists there, although actually doing it might be complicated.

> 	A tuple, strictly speaking, is not *exactly* the same as a list in python.  What you might have happen is this:

>

> 	Command(XXX, YYY, (AAA, BBB))

>

> gets translated by layers below as

>

> 	Command(XXX, YYY, [(AAA,BBB)])

>

> i.e., that tuple becomes a singleton dictionary with a single key-value pair, rather than a list with two elements.

>

> 	Thus, I would disagree, looking at it from a strictly "pythonic" view, that it violates POLA.  I would actually expect it not to work properly.

>

Likely the third argument should allow any iterable.
So it could be a generator even.
The contents of a single element would be what you'd then have to check 
to see if they matched what was expected. (To address your [(aaa,bbb)] 
above.)


-Bill