<
Corbin>
[translation:ERROR] AttributeError: 'SomeInstance' object has no attribute 'no_nul'
<
Corbin>
What did I do wrong? I don't think I've seen this before.
<
Corbin>
Found it. I was trying to use str.join on a list of non-strings.
<
mattip>
pypy v7.3.8 is released. pypy.org/blog and doc.pypy.org have the latest release annoucement
lritter has joined #pypy
greedom has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
jacob22_ has quit [Quit: Konversation terminated!]
<
cfbolz>
mattip: woo, congratulations Matti
otisolsen70 has joined #pypy
<
mgorny>
hm, i wonder if i'm going to OOM translating 6 variants of pypy in parallel
<
cfbolz>
mgorny: maybe during the C compiler phase
<
cfbolz>
Also, depends on how much memory you have 😉
<
mgorny>
but 2 builds are i386, so they'll eat less
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
nimaje has quit [Quit: WeeChat 3.4]
nimaje has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
<
mgorny>
looks like we have test regressions in pypy2.7 final
<
mgorny>
FAIL lib-python/2.7/test/test_xml_etree.py::unmodified
<
mgorny>
FAIL lib-python/2.7/test/test_minidom.py::unmodified
<
mgorny>
FAIL lib-python/2.7/test/test_xml_etree_c.py::unmodified
<
mgorny>
hmm, unless it's due to expat upgrade
<
mgorny>
i'm going to test more
<
mgorny>
yes, it's due to expat upgrade ;-/
<
mgorny>
let's see if cpython is affected
<
mgorny>
yes, cpython also fails
<
mgorny>
xml.etree.ElementTree.XML('<prefix:localname xmlns:prefix="${stuff}"/>')
<
mgorny>
apparently this is no longer considered valid XML
<
mgorny>
pyexpat.ExpatError: syntax error: line 1, column 0
<
LarstiQ>
mgorny: but why?
<
mgorny>
+ #561 CVE-2022-25236 -- Passing (one or more) namespace separator
<
mgorny>
+ characters in "xmlns[:prefix]" attribute values
<
mgorny>
i guess this one
<
mgorny>
i'm going to check cpython git now
<
mgorny>
it just sucks it coincided with pypy final
otisolsen70_ has joined #pypy
otisolsen70 has quit [Ping timeout: 272 seconds]
<
cfbolz>
ok, so I landed assert rewriting in apptest_*.py in py3* branches
<
cfbolz>
huge quality of life improvement for doing py3 work for me :-)
<
mgorny>
let's hope they're going to take 2.4.5 more seriously than they did 2.4.4
slav0nic has joined #pypy
lritter has quit [Ping timeout: 240 seconds]
<
mattip>
I reset the buildbot to skip the py3.7 automatic nightly builds and benchmark runs
<
mattip>
it can still be triggered manually if we desire
otisolsen70__ has joined #pypy
otisolsen70_ has quit [Ping timeout: 272 seconds]
greedom has quit [Remote host closed the connection]
<
Dejan>
congrats on the release
<
LarstiQ>
is the beta removal happening with this release or the next?
<
LarstiQ>
or should that be 3.9 being beta
<
mattip>
whoops, bad copy-pasta. I need a copy-editor. Fixing, thanks
greedom has joined #pypy
<
mattip>
LarstiQ: fixed, thanks
<
LarstiQ>
mattip: np, thanks for shepherding the release!
Atque has quit [Ping timeout: 240 seconds]
<
mattip>
mgorny: the libexpat we ship in our portable builds seems to be expat_2.1.0, from centos7
<
mattip>
windows is expat_2.2.4
Atque has joined #pypy
<
mgorny>
mattip: well, anything below 2.2.5 is vulnerable, and these old versions are scary
<
mgorny>
(but 2.2.5 is buggy, they've just released 2.2.6)
<
mgorny>
2.4.5*/2.4.6*
<
mattip>
I should check all the dependencies before a release I guess
<
mgorny>
2.4.5 alone fixed 5 CVEs
<
mattip>
there is sqlite3 too, which we could have updated
<
mgorny>
well, fortunately distros build against shared libs, so not that many people are affected
<
mgorny>
i guess you could do a .post1 "portable only" release
otisolsen70__ has quit [Ping timeout: 240 seconds]
otisolsen70__ has joined #pypy
otisolsen70__ has quit [Ping timeout: 256 seconds]
greedom has quit [Remote host closed the connection]
epony has quit [Ping timeout: 240 seconds]
greedom has joined #pypy
epony has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
<
mattip>
in devmode, cpython3.9 checks the encoding/decoding errors lookup and raises on this
<
mattip>
python3.9 -X dev -c "bytes('', encoding='ascii', errors='Boom')"
<
mattip>
LookupError: unknown error handler name 'Boom'
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
<
mgorny>
(with a chance for pull request)
otisolsen70 has joined #pypy
otisolsen70_ has joined #pypy
otisolsen70_ has quit [Remote host closed the connection]
otisolsen70 has quit [Ping timeout: 256 seconds]
otisolsen70 has joined #pypy
<
mgorny>
i don't mind making pypy2.7 required for that; i'd just like a confirmation if it's not intended to be supported
<
mattip>
"File "/usr/lib64/python2.7/sre_compile.py", line 583, in compile, "sorry, but this version only supports 100 named groups"
<
mgorny>
yep, somewhat scary ;-)
<
mgorny>
i can't think of a way around this tbh
<
mattip>
the weird thing is pypy2.7 also has this limitation
<
mgorny>
ok, now that's weird
<
mgorny>
do you want me to try printf-debugging what regexp it's compiling?
<
mattip>
maybe cfbolz will see this tomorrow, it is quite late here already
<
cfbolz>
It comes from cffi
<
cfbolz>
Sorry, I'm on my way to bed though
<
mgorny>
i'm sitting up late today
<
mgorny>
slept a lot yesterday and today morning because of no electricity
<
mgorny>
and another storm is expected tonight, so probably no electricity tomorrow again
<
mattip>
it seems I forgot to add PYPY_MAKE_PORTABLE=1 to the aarch64 buildbot, so the package is not portable
otisolsen70 has quit [Quit: Leaving]
<
mattip>
I think I can upload a package with the same release but with a different tag,
<
mattip>
like pypy3.9-v7.3.8-aarch64-rel2.tar.bz2
<
mattip>
or pypy3.9-v7.3.8-aarch64-portable.tar.bz2
<
mgorny>
well, at a first glance it seems that when translated with pypy, it generates more smaller regexes
<
mgorny>
yep, python2.7 seems to get one huge regex
<
mgorny>
while pypy2.7 gets it split into smaller regexes
<
mgorny>
e.g. there's one regex from t_PPHASH to p_INT_CONST_BIN, then another from t_BAD_CONST_OCT...
<
mgorny>
while with cpython it just lands in one regex
<
mgorny>
sorry, it's too much work to figure this out ;-/
<
mgorny>
apparently there's some magical difference in how ply's lexer behaves
greedom has quit [Remote host closed the connection]
slav0nic has quit [Ping timeout: 240 seconds]
Atque has quit [Quit: ...]