Message202039
| Author |
tim.golden |
| Recipients |
astrand, cvrebert, debatem1, gregory.p.smith, ncoghlan, ned.deily, pitrou, rosslagerwall, tim.golden, vstinner |
| Date |
2013年11月03日.17:19:21 |
| SpamBayes Score |
-1.0 |
| Marked as misclassified |
Yes |
| Message-id |
<1383499161.51.0.311586658377.issue9922@psf.upfronthosting.co.za> |
| In-reply-to |
| Content |
The code I've just committed to issue10197 means that the get[status]output functions now pass their (few) tests on all platforms. More by chance than judgement, that change employed universal_newlines which has the effect of forcing the output of check_output to string rather than bytes.
Having just re-read all the comments, we have three positions:
a) Do nothing: these are outdated functions and anyone who has a problem with undecodable bytes will have to use one of the other functions where they have more control.
b) Use the surrogateescape encoding in every case to produce *some* kind of output rather than an exception.
c) Tweak the code to produce bytes in every case. This is actually simple: removing universal_newlines support will do this. (I already have working code for this).
I think (b) is more trouble than it's worth. So (a) Do Nothing; or (c) Produce Bytes.
Going by the law of "Status Quo wins", if no-one chimes in with a strong case for switching to bytes in a few days, I propose to close this as Won't Fix. |
|