Author ncoghlan
Recipients eric.snow, htgoebel, ncoghlan, vstinner
Date 2018-03-12.12:28:27
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1520857707.85.0.467229070634.issue33042@psf.upfronthosting.co.za>
In-reply-to
Content
I don't think we made any start-up changes that were specific to PySys_AddWarnOption, so my suspicion is that the crash is going to be related to a change in the constraints on either the unicode object creation or the list append operation.

The attached patch adds a new test case (I also cleaned up several other details related to test_embed execution in order to more easily see where the segfault happens on failure).

Unfortunately, it isn't yet suitable for use with `git bisect`, as I checked 3.7.0a3 (the earliest tag where the patch applied cleanly), and that commit also segfaults. So setting up for git bisect testing (as per the hot-fix example in https://git-scm.com/docs/git-bisect#_examples ) will require:

* checking the commit where "test_embed.py" was extracted from "test_capi.py" and seeing whether or not that segfaults
* if it *doesn't* segfault, git bisect between there and v3.7.0a3
* if it *does* segfault, make a revised test patch that applies cleanly to early versions, and then go back through the 3.7.0 pre-releases to find one that works
History
Date User Action Args
2018-03-12 12:28:28ncoghlansetrecipients: + ncoghlan, htgoebel, vstinner, eric.snow
2018-03-12 12:28:27ncoghlansetmessageid: <1520857707.85.0.467229070634.issue33042@psf.upfronthosting.co.za>
2018-03-12 12:28:27ncoghlanlinkissue33042 messages
2018-03-12 12:28:27ncoghlancreate