This is a regression introduced in 9211a5d7d0b4, when uses of ST_MTIME
constants were changed to uses of st_mtime attributes. As diagnosed in
the bug report, this change is not merely stylistic: st_mtime is a
float but ST_MTIME’s resolution is rounded to the seconds, so there was
a mismatch between the values seen by file_util and dep_util which
caused an sdist to be unnecessarily created a second time on an ext4
filesystem.
This patch has been tested by John S. Gruber, who reported the bug.
As this is a simple code revert, I think it’s okay to commit without a
unit test.
The changed behavior of sdist in 2.7 broke packaging for projects that
wanted to use a manually-maintained MANIFEST file (instead of having a
MANIFEST.in template and letting distutils generate the MANIFEST).
The fixes that were committed for #8688 (d29399100973 by Tarek and
f7639dcdffc3 by me) did not fix all issues exposed in the bug report,
and also added one problem: the MANIFEST file format gained comments,
but the read_manifest method was not updated to handle (i.e. ignore)
them. This changeset should fix everything; the tests have been
expanded and I successfully tested with Mercurial, which suffered from
this regression.
I have grouped the versionchanged directives for these bugs in one place
and added micro version numbers to help users know the quirks of the
exact version they’re using. I also removed a stanza in the docs that
was forgotten in Tarek’s first changeset.
Initial report, thorough diagnosis and patch by John Dennis, further
work on the patch by Stephen Thorne, and a few edits and additions by
me.
* ftpwrapper now uses reference counting to ensure that the underlying socket
is closed when the ftpwrapper object is no longer in use
* ftplib.FTP.ntransfercmd() now closes the socket if an error occurs
Initial patch by Victor Stinner.
The RFC doesn't say that they are allowed; apparently many mailers accept
them, but not postfix. Contributions to this patch were made by Felipe Cruz
and Catalin Iacob.
cleared by the garbage collector. This fixes a segfault when an instance
and its type get caught in a reference cycle, and the instance's
deallocator calls one of the methods on the type (e.g. when subclassing
IOBase).
Diagnosis and patch by Davide Rizzo.