Author ncoghlan
Recipients emilyemorehouse, eric.snow, hroncok, ncoghlan, vstinner
Date 2018-06-30.11:10:30
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1530357030.84.0.56676864532.issue34008@psf.upfronthosting.co.za>
In-reply-to
Content
At the very least, I think this calls for a documentation change, such that we make it clear that:

1. The expected calling pattern is to call Py_Main() *instead of* Py_Initialize(), since Py_Main() calls Py_Initialize().

2. If you do call Py_Initialize() first, then exactly which settings that Py_Main() will read from the environment and apply is version dependent.

So I'll put together a docs patch that's valid for 3.6+, and then we can decide where to take 3.7 and 3.8 from an implementation perspective after that.

Currently, I'm thinking that for 3.7, our goal should specifically be "fake the 3.6 behaviour well enough to get fontforge working again" (so it will mostly be Fedora that drives what that 3.7.1 change looks like), while for 3.8 we should aim to deprecate that code path, and instead offer a better building block API for folks that want to implement their own Py_Main() variant.
History
Date User Action Args
2018-06-30 11:10:30ncoghlansetrecipients: + ncoghlan, vstinner, eric.snow, emilyemorehouse, hroncok
2018-06-30 11:10:30ncoghlansetmessageid: <1530357030.84.0.56676864532.issue34008@psf.upfronthosting.co.za>
2018-06-30 11:10:30ncoghlanlinkissue34008 messages
2018-06-30 11:10:30ncoghlancreate