Decode log messages on windows as UTF-8 #153
Conversation
1 similar comment
| # subprocess.Popen defaults to locale.getpreferredencoding() | ||
| # which can be cp1252 (for example) | ||
| if (decode_text and platform.system() == "Windows" | ||
| and sys.version_info >= (3, 6)): |
There was a problem hiding this comment.
This version seems arbitrary. It's not good practice to hard-code specific versions. If absolutely necessary, it's better to test for the features/characteristics that you actually need. In this case, you can just pass universal_newlines=True, which is supported in both 2.7 and 3.x .
There was a problem hiding this comment.
Since we're already passing universal_newlines (
Line 39 in e8a482d
string in Python 3.x in situations where text is expected. This pull-request is either in response to a not-understand problem or it's a symptom of a different problem. In which context are you encountering this?
There was a problem hiding this comment.
Please open an issue with your usage and the error that you're encountering and go from there.
|
Closing this PR in lieu of an issue to discuss the problem. See comments above. |
|
Hi, I've raised #154 The version isn't arbitrary, it is the first version at which python allowed an encoding to be passed to Popen rather than selecting the default encoding itself - it isn't available in 3.5 https://docs.python.org/3.5/library/subprocess.html#subprocess.Popen |
On Windows svn (at least for current versions) pipes output encoded as UTF-8. I discovered this by having PySvn incur an exception when attempting to decode the output of log that had non-ascii punctuation characters in it.