Author xdegaye
Recipients Arfrever, brett.cannon, doko, ezio.melotti, martin.panter, nadeem.vawda, petri.lehtinen, python-dev, serhiy.storchaka, vstinner, xdegaye
Date 2017-01-19.16:54:54
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1484844894.71.0.733739713658.issue13886@psf.upfronthosting.co.za>
In-reply-to
Content
With input-readline.v3.patch, test_builtin runs with success on Android api 21.

With pep538_coerce_legacy_c_locale_v4.diff that implements PEP 538 in issue 28180, and with input-readline.v3.patch modified to have 'readline_encoding = locale.getpreferredencoding()' even when 'is_android' is True, test_builtin runs with success. This means that no specific Android handling would be needed if PEP 538 were to be adopted.

The new input-readline.v4.patch is a slight improvement over the previous patch and sets readline_encoding to 'UTF-8' on Android when test_builtin is run with the environment variable LANG set to 'en_US.UTF-8' and in that case the test exercises all the code paths including those with the readline module.  This is because locale.getdefaultlocale() returns ('en_US', 'UTF-8') on Android in that case and because both getdefaultlocale() and readline scan the environment to check for the locale.  This new patch is only useful if tests on Android are expected to be run also with one of the locale environment variables set to UTF-8 (and PEP 538 is rejected).

I have left some comments on Rietveld.
History
Date User Action Args
2017-01-19 16:54:54xdegayesetrecipients: + xdegaye, brett.cannon, doko, vstinner, nadeem.vawda, ezio.melotti, Arfrever, python-dev, petri.lehtinen, martin.panter, serhiy.storchaka
2017-01-19 16:54:54xdegayesetmessageid: <1484844894.71.0.733739713658.issue13886@psf.upfronthosting.co.za>
2017-01-19 16:54:54xdegayelinkissue13886 messages
2017-01-19 16:54:54xdegayecreate