- Tools/Demos
- gh-129248: The iOS test runner now strips the log prefix
from each line output by the test suite.
- gh-104400: Fix several bugs in extraction by switching to
an AST parser in pygettext.
- Tests
- gh-129386: Add test.support.reset_code, which can be used
to reset various bytecode-level optimizations and local
instrumentation for a function.
- gh-128474: Disable test_embed test cases that segfault on
BOLT instrument binaries. The tests are only disabled when
BOLT is enabled.
- gh-128003: Add an option --parallel-threads=N to the
regression test runner that runs individual tests in
multiple threads in parallel in order to find concurrency
bugs. Note that most of the test suite is not yet reviewed
for thread-safety or annotated with @thread_unsafe when
necessary.
- Security
- gh-105704: When using urllib.parse.urlsplit() and
urllib.parse.urlparse() host parsing would not reject
domain names containing square brackets ([ and ]). Square
brackets are only valid for IPv6 and IPvFuture hosts
according to RFC 3986 Section 3.2.2.
- gh-126108: Fix a possible NULL pointer dereference in
PySys_AddWarnOptionUnicode().
- gh-80222: Fix bug in the folding of quoted strings
when flattening an email message using a modern email
policy. Previously when a quoted string was folded so
OBS-URL: https://build.opensuse.org/package/show/devel:languages:python:Factory/python314?expand=0&rev=38
- Tools/Demos
- gh-132121: Always escape non-printable Unicode characters in
pygettext.
- gh-131852: msgfmt no longer adds the POT-Creation-Date to
generated .mo files for consistency with GNU msgfmt.
- Tests
- gh-131277: Allow to unset one or more environment variables
at once via EnvironmentVarGuard.unset(). Patch by Bénédikt
Tran.
- gh-131050: test_ssl.test_dh_params is skipped if the
underlying TLS library does not support finite-field
ephemeral Diffie-Hellman.
- Security
- gh-131809: Update bundled libexpat to 2.7.1
- gh-131261: Upgrade to libexpat 2.7.0
- gh-121284: Fix bug in the folding of rfc2047 encoded-words
when flattening an email message using a modern email
policy. Previously when an encoded-word was too long for
a line, it would be decoded, split across lines, and
re-encoded. But commas and other special characters in the
original text could be left unencoded and unquoted. This
could theoretically be used to spoof header lines using a
carefully constructed encoded-word if the resulting rendered
email was transmitted or re-parsed.
- Library
- gh-132174: Fix function name in error message of
_interpreters.run_string.
- gh-132171: Fix crash of _interpreters.run_string on string
subclasses.
OBS-URL: https://build.opensuse.org/package/show/devel:languages:python:Factory/python314?expand=0&rev=44
