New submission from Larry Hastings:
Right now the atomic access fence macros in pyatomic.h are unconditional. This
means that they're active even even when you "./configure --without-threads".
If Python thread support is disabled, surely we don't need to ensure atomic
Larry Hastings added the comment:
Patch attached. The basics were okay; however, there was no locking around
access to a static variable (_Py_AllocatedBlocks) so I added some. The way the
code managed _Py_AllocatedBlocks was a bit odd; this approach resulted in fewer
lines, but it was hard
Larry Hastings added the comment:
Oy veh, in editing the issue I dropped the attached file. Here it is.
--
keywords: +patch
Added file:
http://bugs.python.org/file42479/larry.refresh.lock.macros.for.obmalloc.diff
___
Python tracker
<h
Larry Hastings added the comment:
Yes, I've been working on a patch to do this as well. I called the calling
convention METH_RAW, to go alongside METH_ZERO METH_O etc. My calling
convention was exactly the same as yours: PyObject *(PyObject *o, PyObject
**stack, int na, int nk). I onl
Larry Hastings added the comment:
I did the path_converter change. IIRC some functions supported bytearray, some
did not, and in my quest for consistency I took the superset of functionality
and supported bytearray for everything.
Sounds to me like bytearray support should be dropped, but
Changes by Larry Hastings :
--
nosy: -larry
___
Python tracker
<http://bugs.python.org/issue26825>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Larry Hastings:
So here's a strange one.
I want to do some mysterious experiments with CPython. So I disabled refcount
changes in CPython.
I changed Py_INCR and Py_DECR so they expand to nothing. I had to change some
other macros to match (SETREF, XSETREF, an
Changes by Larry Hastings :
--
versions: +Python 3.6
___
Python tracker
<http://bugs.python.org/issue26891>
___
___
Python-bugs-list mailing list
Unsubscribe:
Larry Hastings added the comment:
I did as you suggested. I also discovered that _Py_NewReference is usually a
macro, and I'd only fixed the version in Objects/object.c (the one not being
used). When I fixed both those things, and switched the makefile so it uses a
different P
Larry Hastings added the comment:
Here is my theory: if the code asserts that it's true, and the code normally
runs fine, then it's normally true, and therefore I can remove the assertion
and the code will run correctly. I haven't hit that assertion in the peephole
optimiz
Larry Hastings added the comment:
I will admit that even during development I rarely used Clinic's test suite.
My test suite is: make a change, then run Clinic over all of CPython and do a
diff. If nothing changed then Clinic is still wo
Larry Hastings added the comment:
This is caused by your "meth_o inline" code, which isn't checked in yet. A
clean checkout of CPython trunk generates this correct code:
if (!PyArg_Parse(arg, "O&:test", PyUnicode_FSConverter, &path))
Maybe you don'
Larry Hastings added the comment:
I'm not excited by this new addition to PEP 7, so I'm excusing myself from this
issue.
--
___
Python tracker
<http://bugs.python.o
Larry Hastings added the comment:
Why is this dependent on #26305?
--
___
Python tracker
<http://bugs.python.org/issue23867>
___
___
Python-bugs-list mailin
Larry Hastings added the comment:
This has nothing to do with Argument Clinic, which is an internal tool used in
the Python 3.x series.
--
components: -Argument Clinic
nosy: -859911096
___
Python tracker
<http://bugs.python.org/issue26
Larry Hastings added the comment:
Christian: any interest in proposing this for 2.7? We could ask Benjamin. It
could still make 2.7.11--rc1 should be tagged in about a month.
--
___
Python tracker
<http://bugs.python.org/issue16
Larry Hastings added the comment:
Summary: read Matt's original message on the bug.
Status: still broken. Some people think it should be left that way.
Proposed fix: it's straightforward to fix. You wrote a patch, I modified yours
a little. The patch is short and ea
Larry Hastings added the comment:
Updated patch based on feedback from Serihy. Thanks for the ping.
--
Added file:
http://bugs.python.org/file42860/larry.make.file_read.use.size_t.2.diff
___
Python tracker
<http://bugs.python.org/issue21
Larry Hastings added the comment:
Martin said:
> Josh: it's not as simple as just changing the type code and variable type.
> Systems used to require all kinds of types in the length parameter of read(2)
> and fread(3), including int, long, and size_t. If it was int, passing s
Larry Hastings added the comment:
Can you give me a test case?
--
___
Python tracker
<http://bugs.python.org/issue23500>
___
___
Python-bugs-list mailin
Larry Hastings added the comment:
Perhaps, now that Guido allows us to redirect into a separate file, we should
simply abandon the two-pass approach.
--
___
Python tracker
<http://bugs.python.org/issue20
Larry Hastings added the comment:
This can wait for 3.5.
--
___
Python tracker
<http://bugs.python.org/issue19075>
___
___
Python-bugs-list mailing list
Unsub
Larry Hastings added the comment:
This is part of the release process. 3.4.3 was fine last I checked, so perhaps
some wonderful automated process broke it for me. 3.5, I think I simply didn't
upload the docs properly.
Anyway, this is my responsibility as RM for 3.4 and 3.5, so in the f
Larry Hastings added the comment:
I turned in my Windows developer badge in 2007. Can I recuse myself,
pretty-please?
How about Tim Golden or Zach Ware? Who I notice are conveniently already added
to the nosy list!
--
___
Python tracker
<h
Larry Hastings added the comment:
I think I've got them all working now. Please reopen if you discover new
breakage. (Or old breakage I guess if I didn't actually fix it...!)
--
resolution: -> fixed
stage: -> resolved
status
Larry Hastings added the comment:
You give me too much credit. I'm the chimp who put it there in the first
place. Though, admittedly, it was a copy&paste job based on the ancient
CObject.
My guess is this was very helpful in 1.4, or
Larry Hastings added the comment:
Yeah, I went ahead and checked this in. I'm living life on the *edge*.
To the EXTREME!
--
assignee: -> larry
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Pyth
Larry Hastings added the comment:
In your sample output we still get two #ifndef SPAM_METHODDEF stanzas.
Wouldn't it be better to only have one?
Maybe Clinic needs to be smarter about generating those anyway. Let me think
about it.
--
___
P
Larry Hastings added the comment:
I can live with that, if you're confident in it and you'll watch the buildbots.
--
___
Python tracker
<http://bugs.python.o
Larry Hastings added the comment:
FWIW I'm tagging alpha 2 somewhere around 24-36 hours from now.
--
___
Python tracker
<http://bugs.python.org/is
Larry Hastings added the comment:
Using it with 3.5.0a2, it still printed the "could not be processed" error
message, but there's a new wrinkle: the one executable it processed, it added
incorrectly. The "-amd64-webinstall" exe was identified as "Win
Larry Hastings added the comment:
Yup.
--
___
Python tracker
<http://bugs.python.org/issue23619>
___
___
Python-bugs-list mailing list
Unsubscribe:
Larry Hastings added the comment:
> Any idea how long until the CDN picks up the update?
curl -X PURGE https://www.python.org/ftp/python/2.7.5/Python-2.7.5.tar.xz
will explicitly tell the CDN to pick up the new binaries. Not sure if that
works from anywhere or just from dl-files.iad1.psf
Larry Hastings added the comment:
Okay, I refreshed the MD5 sums and sizes on the web site, and purged the old
files from the CDN.
--
___
Python tracker
<http://bugs.python.org/issue23
Larry Hastings added the comment:
My inclination is 3.5 only. Barry, do you want to argue for this going into
3.4?
--
___
Python tracker
<http://bugs.python.org/issue23
Larry Hastings added the comment:
Oh, I read the code. But it's a performance hack, and the rules say we only
accept security fixes and bug fixes at this stage of the release, and they're
the rules for good reasons.
--
___
Python trac
Larry Hastings added the comment:
How about this approach? Only ever emit the #ifndef stanza once per symbol.
--
Added file:
http://bugs.python.org/file38491/larry.ac_multiple_macro_definitions.diff.1.txt
___
Python tracker
<http://bugs.python.
Larry Hastings added the comment:
(see larry.ac_multiple_macro_definitions.diff.1.txt posted above)
--
___
Python tracker
<http://bugs.python.org/issue23
Larry Hastings added the comment:
Oops, I should have run "make clinic", so you could see all the changes that
result from this patch.
--
Added file:
http://bugs.python.org/file38492/larry.ac_multiple_macro_definitions.diff.2.txt
___
Pyth
Larry Hastings added the comment:
What do you think of this approach? Now a "Destination" object behaves like an
array of text accumulators. If you ask for one that doesn't exist it's created
for you. When the Destination is dumped, the output from each accumulator is
j
Larry Hastings added the comment:
I read it quickly. It looks basically okay, but I have one issue to discuss
right now. Changing from the format unit 'O!' and calling "getuchar" to the
format unit 'C' means a change in semantics. The text in the exceptio
Larry Hastings added the comment:
I want to redo it--it's smelly. I hope to get it done this week.
--
___
Python tracker
<http://bugs.python.org/is
Larry Hastings added the comment:
As promised, here's a cleaned-up version of the patch. The taxonomy of objects
now makes sense: a Destination contains a BufferSeries object, rather than
Destinations weirdly supporting __getitem__ to reference different objects.
I tripped over mys
Larry Hastings added the comment:
I did, I just didn't respond. I'll do that now.
--
___
Python tracker
<http://bugs.python.org/issue23500>
___
___
Larry Hastings added the comment:
Updated patch, removed all references to two-pass. Also realized I needed to
make the default behavior for methoddef_ifndef go to the end. And, finally, I
forgot to merge the "only print each #ifndef block once" code I wrote before
when I redid
Larry Hastings added the comment:
Does this really need a backport to 3.4?
--
___
Python tracker
<http://bugs.python.org/issue23500>
___
___
Python-bugs-list m
Larry Hastings added the comment:
Removing 3.4 from the version list as I close the bug, then. If we decide we
need it backported please reopen (or create a new bug, either is fine).
--
resolution: -> fixed
stage: commit review -> resolved
status: open -> closed
versions:
Larry Hastings added the comment:
Thanks! I've been meaning to follow up on this. I think CPython should be
consistent, one way or another. So I support this change.
--
___
Python tracker
<http://bugs.python.org/is
New submission from Larry Hastings:
I was staring at Clinic tonight and a thought came to me. We could express
"nullable=True" as "types='NoneType'". For example, the converter for 'Z'
could change from
Py_UNICODE(nullable=True)
to
Py_UNICOD
Larry Hastings added the comment:
Saying that "str" is redundant makes me think you don't understand what "types"
does. "types" accepts a text string listing all of the Python types the
converter accepts. You can accept the converter's default, or
New submission from Larry Hastings:
tl;dr: skipitem() in Python 3 still supports "w" and "w#", which were
abandoned. They should be removed.
--
If:
* you have an extension that parses its arguments with
PyArg_ParseTupleAndKeywords (or its cousins),
* you have
New submission from Larry Hastings:
tl;dr: skipitem() in Python/getsargs.c doesn't know about the 'w*' format unit,
which could lead to a crash. It should know about it.
--
If:
* you have an extension that parses its arguments with
PyArg_ParseTupleAndKeywords (or its cousi
Larry Hastings added the comment:
Would this qualify as an "easy" bug we could throw to a sprinter?
--
___
Python tracker
<http://bugs.python.o
Larry Hastings added the comment:
I finally talked to someone (Thomas Wouters) about this. For something going
into trunk that will be part of 3.5, this is fine.
Please include a mention that the exception's text messages changed in
Misc/NEWS. I can help you write this if you want.
New submission from Larry Hastings:
Clinic was previously pretty confused about what things accepted "bytes",
"bytearray", "buffer", "robuffer", and "rwbuffer". This is because the
documentation itself was somewhat confusing.
The docum
Larry Hastings added the comment:
Huh. Why didn't it attach my patch? Here it is.
--
Added file:
http://bugs.python.org/file38933/larry.clinic.buffer.conceptual.cleanup.1.txt
___
Python tracker
<http://bugs.python.org/is
Larry Hastings added the comment:
LGTM, just adding Antoine (the expert for the socket module). I'm ready to
commit over here.
--
nosy: +pitrou
___
Python tracker
<http://bugs.python.org/is
Larry Hastings added the comment:
Updated the patch a little, to make it enforce compatible "types" parameters.
The previous patch had some functions that would permit passing in "types='blah
de blah'" or other nonsense.
--
Added file:
ht
Larry Hastings added the comment:
Follow the instructions in this:
https://docs.python.org/devguide/docquality.html
And make it look something like this:
https://docs.python.org/3/library/socket.html#socket.CAN_BCM
--
___
Python tracker
<h
Larry Hastings added the comment:
Patch committed. Thanks, Joe!
--
assignee: -> larry
resolution: -> fixed
stage: commit review -> resolved
status: open -> closed
___
Python tracker
<http://bugs.python
Larry Hastings added the comment:
LGTM
--
assignee: -> zach.ware
___
Python tracker
<http://bugs.python.org/issue20586>
___
___
Python-bugs-list mai
Larry Hastings added the comment:
If you give me a test case I'll knock this out.
--
___
Python tracker
<http://bugs.python.org/issue23944>
___
___
Pytho
Larry Hastings added the comment:
Isn't this a "new feature"? 2.7 is maybe fair game, but 3.4 arguably is not.
It should be odd to have a field marked in the documentation as "New in 3.4.4."
--
___
Python tracker
<http:
Larry Hastings added the comment:
Benjamin, would you like CAN_RAW_FD_FRAMES in the next 2.7?
--
nosy: +benjamin.peterson
___
Python tracker
<http://bugs.python.org/issue22
New submission from Larry Hastings:
My laptop is running 64-bit Linux (14.10). It has 4.6GB of free disk space.
Naturally that's not enough to run test_mmap.
When I run the test suite, test_mmap consumes all available disk space, then
fails. (Hopefully freeing all its temporary files!
Larry Hastings added the comment:
New patch revision, including the new API change (the "types" argument to a
constructor must now be a set of strings).
--
Added file:
http://bugs.python.org/file39001/larry.clinic.buffer.conceptual.cle
Larry Hastings added the comment:
I wish the Python bug tracker had a way of attaching files to issues, so we
didn't have to just copy and paste into and out of the comments. Ah well,
wishful thinking!
--
___
Python tracker
Larry Hastings added the comment:
That's AMAZING, you mean the web site knew how to do that THE WHOLE TIME?
--
___
Python tracker
<http://bugs.python.org/is
Larry Hastings added the comment:
Attached is a patch. This causes a lot of churn, darn.
Should I copy & paste the patch into this text box too? Maybe that would be
more convenient!
--
Added file:
http://bugs.python.org/file39008/larry.clinic.wrap.long.impl.decls.1
Larry Hastings added the comment:
Okay, this patch is about half the size. I had a bug that was throwing away
some important syntax. With this patch applied, Python builds and the test
suite is as reasonable as usual.
I swear, C compilers are so fussy sometimes!
--
Added file
Changes by Larry Hastings :
--
assignee: -> larry
resolution: -> fixed
stage: needs patch -> resolved
status: open -> closed
___
Python tracker
<http://bugs.python
Larry Hastings added the comment:
ZFS.
--
___
Python tracker
<http://bugs.python.org/issue23953>
___
___
Python-bugs-list mailing list
Unsubscribe:
Larry Hastings added the comment:
Wall time was 3 minutes 38 seconds.
% time ./python -m test test_mmap
[1/1] test_mmap
test test_mmap failed -- multiple errors occurred; run in verbose mode for
details
1 test failed:
test_mmap
./python -m test test_mmap 0.40s user 22.41s system 10% cpu 3
Larry Hastings added the comment:
In the case of namedtuple and Enum, the parameter represents a sequence of
strings--order is significant. With the 'types' parameter for converters, the
internal model was always meant to be a *set* of strings. The order was
explicitly *not* s
Larry Hastings added the comment:
Latest patch, with another round of lovely comments from Serhiy.
--
Added file:
http://bugs.python.org/file39030/larry.clinic.buffer.conceptual.cleanup.4.txt
___
Python tracker
<http://bugs.python.org/issue23
Larry Hastings added the comment:
If
* the diff looks clean
* it compiles without any *new* (sigh) errors, and
* it passes the unit test suite without any *new* (sigh) failures,
then the Clinic conversion can generally be considered a success
New submission from Larry Hastings:
Peter's working on converting socket to use Argument Clinic. He had a default
that really should look like this:
min(SOME_SOCKET_MODULE_CONSTANT, 128)
"min" wasn't something we'd needed before. I thought about it and rea
New submission from Larry Hastings:
Matt Mackall brought this over to me today.
If list "l" has one million pairs and you do
dict(l)
it creates the dict, then resizes it to a million elements, then starts adding
name/value pairs. But since it's sized to a million elements,
Larry Hastings added the comment:
While cleaning this up I noticed that es# and et# should both require
"nullable". Fixed that too. I sure hope it's all correct now!
--
resolution: -> fixed
stage: patch review -> resolved
s
New submission from Larry Hastings:
Documentation is here:
https://docs.python.org/3/c-api/arg.html#arg-parsing
The first line of documentation for each format unit follows this convention:
formatunit (pythontype) [arguments, to, pyarg_parsetuple]
These represent the format unit
Larry Hastings added the comment:
> The "e" variants (typically) allocate a buffer for you, since it's pretty
> much unknown how long the encoded data will be.
All four will do it if you pass in a NULL pointer. "es#" and "et#" can reuse
an existing buff
Larry Hastings added the comment:
One more idea. We annotate with an & when you pass in a pointer to a variable.
So format unit 'i' would get [& int], 's' would get [& char *], and 'es#'
would get [char *, & char *, & int].
--
New submission from Larry Hastings:
I found another bug in the mapping of converters to format units. (s#, z#, and
y# all allow zeroes.)
I've redone the approach for str_converter in an attempt to make it easier to
read.
It'd be swell if, after this gets checked in (or rejected)
New submission from Larry Hastings:
New proposed semantics for the types= parameter to converters: where possible,
pass in actual types. The resulting syntax:
c: int(types={str}) # maps to 'U'
s: str(types={str, robuffer}, length=True, zeroes=True) # maps to 's#&
Larry Hastings added the comment:
I'm now converging on changing types= to accept a set of real types, see issue
#24001. That change has an impact on this decision.
(By the way, let's assume that if "nullable", I have to rename it to
"accepts_none". I'll u
Larry Hastings added the comment:
Should types= be renamed accept= ? It's a set of the types of the Python
objects that this parameter should accept.
--
___
Python tracker
<http://bugs.python.org/is
New submission from Larry Hastings:
Twice recently I've wanted a function that transforms an AST node tree back
into text:
* In the hacked-up Tools/clinic/clinic.py for issue #24001
* In the hacked-up Lib/inspect.py for issue #23967
Both times I did a half-assed job just to get the
Larry Hastings added the comment:
I should mention that evalify_node() is pretty hacked up here, and is not ready
to be checked in. (I'm proposing separately that we simply add something like
this directly into the standard library, see issue #
Larry Hastings added the comment:
I should mention that evalify_node() is pretty hacked up here, and is not ready
to be checked in. (I'm proposing separately that we simply add something like
this directly into the standard library, see issue #
Changes by Larry Hastings :
--
nosy: +serhiy.storchaka, zach.ware
___
Python tracker
<http://bugs.python.org/issue24002>
___
___
Python-bugs-list mailin
Larry Hastings added the comment:
Good idea, I'll go ahead and borrow Guido's time machine.
https://docs.python.org/3/library/ast.html#ast.NodeVisitor
However, NodeVisitor does not transform the ast tree back into text. So in
what way is this helpful?
Also, for what it's wor
Larry Hastings added the comment:
Oh, I misremembered. The name "allow_none" was inflicted on me by python-dev,
in this thread:
https://mail.python.org/pipermail/python-dev/2014-August/135650.html
--
___
Python tracker
<http://bu
Larry Hastings added the comment:
Actually eval(ast) is all I need for #23967 too. But eval is a builtin, so it
feels wrong to have it supporting--and therefore dependent on--ast.
--
___
Python tracker
<http://bugs.python.org/issue24
Larry Hastings added the comment:
Sadly, for political reasons, it's best that we not convert collections,
itertools, or random for now. I'll update this issue if the situation changes.
Sorry about that!
--
___
Python trac
Larry Hastings added the comment:
Oh, heavens, yes, that's much nicer. Thanks for the suggestion, Serhiy!
--
Added file:
http://bugs.python.org/file39115/larry.one.more.clinic.format.unit.map.cleanup.2.txt
___
Python tracker
Larry Hastings added the comment:
p.s. I'm now leaning heavily towards renaming "types" to "accept", and putting
NoneType in the set passed to accept= instead of a separate parameter.
--
___
Python tracker
<http:
Larry Hastings added the comment:
I've posted about this to python-dev.
--
___
Python tracker
<http://bugs.python.org/issue23920>
___
___
Python-bugs-list m
Larry Hastings added the comment:
> Doesn't always zeroes == length?
"zeroes" requires "length", but "length" does not require "zeroes". As it
happens all the format units supported by str always have both parameters
either True or False. B
Larry Hastings added the comment:
There is! compile() will do it, though the docstring doesn't mention it. (The
full documentation does.)
The following function meets both my use cases:
def eval_ast_expr(node, symbols=None, *, filename='-'):
"""
Larry Hastings added the comment:
Thanks to #24002 I now know how to write evalify_node properly. This revision
of the patch is much better, and maybe ready for checkin.
--
Added file:
http://bugs.python.org/file39122/larry.one.more.clinic.format.unit.map.cleanup.2.txt
2201 - 2300 of 2389 matches
Mail list logo