[Python-Dev] Late Python 3.7.1 changes to fix the C locale coercion (PEP 538) implementation

[Victor Stinner]

Le mer. 19 sept. 2018 à 09:50, Nick Coghlan <ncoghlan at gmail.com> a écrit :
> I think the changes to both master and the 3.7 branch should be reverted.

Ok, I prepared a PR to revert the 3.7 change:
https://github.com/python/cpython/pull/9416


> For 3.7, I already said that I think we should just accept that that
> ship has sailed with 3.7.0 and leave the as-shipped implementation
> alone for the rest of the 3.7 series: (...) For 3.8, (...), my PR
> should be conflict free again, and we'll be able to get PEP 538 back
> to working the way it was always supposed to work (...)

I read all your comments, and honestly, I don't understand you. Once you say:

"we don't actually want anyone turning off locale coercion except for
debugging purposes"
https://bugs.python.org/issue34589#msg325554

but you also say that Python 3.7.0 is broken on Centos 7 because it's
not possible to disable C locale coercion using -E flag:
https://bugs.python.org/issue34589#msg325246

And here (your email), one more time, you insist to support
"PYTHONCOERCECLOCALE=0 python3 -E".

I don't understand if you want PYTHONCOERCECLOCALE to be ignored when
using -E or not.

Since the PEP 538 is something new, we don't have much feedback of
users to know if it causes any troubles, so I agree that we should
provide a way to disable the feature, as I provided a way to disable
the UTF-8 Mode when the LC_CTYPE is C or POSIX. Just to give user a
full control on locales and encodings.

That's why I came up with a new -X coerce_c_locale option which can be
used even with -E. I understood that you like the option, since you
proposed to use it:
https://bugs.python.org/issue34589#msg325493

--

Moreover, you asked me to make sure that Py_Initialize() and Py_Main()
cannot enable C locale coercion. That's what I did.

--

IMHO the implementation is really a secondary concern here, the main
question is: what is the correct behavior?

Nick:

* Do we agree that we need to provide a way to disable C locale
coercion (PEP 538) even when -E is used?
* Do you agree that Py_Initialize() and Py_Main() must not enable the
C locale coercion (PEP 538)?

I understood that your reply is yes for the second question, since you
insist to push your change which also prevent Py_Initialize() and
Py_Main() to enable C locale coercion.

If we change 3.7.0 behavior in 3.8, I would prefer to change the
behavior in 3.7.1. IMHO it's not too late to *fix* 3.7.

--

I decided to push a concrete implementation because I understood that
you was ok for the -X coerce_c_locale option and you asked me to fix
my mistakes. I feel guilty that I broke the implementation of your PEP
:-( Moreover, I'm also exhausted by fixing locales and encodings, I'm
doing that for one year now, and I expected many times that I was done
with all regressions and corner cases...

We are discussing these issues since 3 weeks and we failed to fix
them, whereas Ned asked to push last fixes before 3.7.1. I sent an
email to make sure that we all agree on the solution.

Well, it seems like again, we failed to agree on the expected *behavior*.

Victor