PEP 308: Alternative conditional operator forms -- Corner Case solved

[Michele Simionato]

"Samuele Pedroni" <pedronis at bluewin.ch> wrote in message news:<3e496778_2 at news.bluewin.ch>...
> "Samuele Pedroni" <pedronis at bluewin.ch> ha scritto nel messaggio
> news:3e4951cc$1_2 at news.bluewin.ch...
> >
> > brilliant, you've just made the problem worse!
> 
> AK> How so?
> 
> You consider having a conditional expression looking like the statement a
> feature and you have worked to _technically_ solve the parser, language
> definition details to have it so...
> 
> I consider that possibility a mixed bless, I would rather (if we should have
> one) have a conditional operator that looks operatorial and in the same
> family
> of 'and'/'or'. Why?
> 
> Because the fact that Python distinguishes statements and expressions is
> basically, apart from being the status quo and what is implemented, a
> _deliberate_ decision,  that happens to work well together with
> indentation-delimited blocks.
> 
> [OTOH is probably not impossible to conceive a language that mixes
> indentation
> usage and () usage as delimiters and where everything is an expression and
> produces a value]
> 
> Having a conditional expression that looks like a statement will touch this
> very nerve ...
> and will blur the distinction from the hand-on experience of the language
> (of
> course not from the lang ref POV).
> 
> I prefer not to be confronted (if possible) with the questions:
> 
> "Why this does not work?" with this being:
> 
> if x: (print x)
> (for x in l: print l)
> ...
> if x>0: print "pos" else: print "negative"
> 
> they make no sense from the lang  ref/lawyer perspective. But if the asker
> is
> naive/confused/a troll/impertinent:
> 
>  answering ye we have this two things that look the same but one is a
> statement
> and one is an expression and we disambiguate them so, put in  [clever
> explanation], but we are for statement/expression distinction and RTFM
> 
> is a kind of "cheap bastard" attitude, because the fact that the above are
> invalid is also a matter of _deliberate_choice, so we would have to go
> philosophical and discuss this is "pythonic" and this not and the
> statement/expression distinction thing, and argue about readability ...
> 
> so if possible I would rather avoid this, thanks.

I must say that even if initially I said "if C: x else: y" is ugly but
better than nothing, in the last few days the arguments of Samuele and
others finally convinced me that "nothing" is better than such a
confusion between expressions and statement. It seems to me absolutely
contrary to the Python design philosophy.
Guido seems to be accepting this solution (see the new form of
the PEP), nevertheless I reserve to myself the right to disagree.

                                Michele
right to di