From arigo at tunes.org  Tue Jun  3 09:56:31 2014
From: arigo at tunes.org (Armin Rigo)
Date: Tue, 3 Jun 2014 09:56:31 +0200
Subject: [pypy-dev] europython sprints?
In-Reply-To: <CAH88djX31ERdmg=J_hEydHwg+_jxez7g1VHpH2gLueWJJ5vr=A@mail.gmail.com>
References: <CAH88djX31ERdmg=J_hEydHwg+_jxez7g1VHpH2gLueWJJ5vr=A@mail.gmail.com>
Message-ID: <CAMSv6X0fa6fYVfMFC_e7BqOHrz+dSuTwsbPm0gXimTKmHaNfJA@mail.gmail.com>

Hi Martin,

On 28 May 2014 19:15, Martin Matusiak <numerodix at gmail.com> wrote:
> I'm wondering if there will be pypy sprints at Europython this year
> (?) And if so, which are the principal topics (I assume stm is one of
> them..)?

As you might have guessed from the lack of answer, we don't know yet.
I guess there will be a PyPy sprint, yes.


A bient?t,

Armin.

From arigo at tunes.org  Tue Jun  3 10:00:17 2014
From: arigo at tunes.org (Armin Rigo)
Date: Tue, 3 Jun 2014 10:00:17 +0200
Subject: [pypy-dev] Getting involved
In-Reply-To: <5386286A.9010300@oracle.com>
References: <5386286A.9010300@oracle.com>
Message-ID: <CAMSv6X1HneEberxhnBUn7m03r6PNoRbMvQNLsGa7_u9EE1_sbA@mail.gmail.com>

Hi Travis,

On 28 May 2014 20:18, Travis Jensen <travis.jensen at oracle.com> wrote:
> I've been following pypy for a while and have thought it would be fun to get
> involved.  Recently, I've been wondering what python with tail call
> optimization would look like (I'm disgruntled by GVR's decision :), and
> realized this might be a good way for me to get involved in pypy.

About this topic, we stick with Guido's language design and don't want
a PyPy in which a number of frames may be skipped from tracebacks.
Now, if you don't want to change the official PyPy but just play
around for fun, then yes, PyPy is a good choice.  You'd have to tweak
the interpreter, around the files interpreter/pyframe.py and
interpreter/pyopcode.py, as well as the bytecode compiler in
interpreter/astcompiler/.


A bient?t,

Armin.

From matti.picus at gmail.com  Fri Jun  6 15:49:04 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Fri, 06 Jun 2014 16:49:04 +0300
Subject: [pypy-dev] release 2.3.1
Message-ID: <5391C6D0.9080404@gmail.com>

Please download and try out the release candidate 71938-32f35069a16d from
http://buildbot.pypy.org/nightly/release-2.3.x
Comments on the release notice
http://doc.pypy.org/en/latest/release-2.3.1.html
are welcome as well. Barring last minute failures, I intend to release 
over the weekend.
Matti


From arigo at tunes.org  Sun Jun  8 09:08:41 2014
From: arigo at tunes.org (Armin Rigo)
Date: Sun, 8 Jun 2014 09:08:41 +0200
Subject: [pypy-dev] PyPy 2.3.1 released
Message-ID: <CAMSv6X2=Hug6+Xg82irYP5gj2Z80RPNJa1=vt+9j+uQTRLq80A@mail.gmail.com>

Hi all,

We're pleased to announce PyPy 2.3.1, a feature-and-bugfix improvement
over our recent 2.3 release last month.


This release contains several bugfixes and enhancements among the
user-facing improvements:

* The built-in struct module was renamed to _struct, solving issues
with IDLE and other modules

* Support for compilation with gcc-4.9

* A CFFI-based version of the gdbm module is now included in our binary bundle

* Many issues were resolved since the 2.3 release on May 8


Complete release notice: http://doc.pypy.org/en/latest/release-2.3.1.html
Download: http://pypy.org/download.html

Please try it out and let us know what you think. We especially
welcome success stories, please tell us about how it has helped you!

Cheers, The PyPy Team (thanks Matti!)

From naylor.b.david at gmail.com  Sun Jun  8 15:22:18 2014
From: naylor.b.david at gmail.com (David Naylor)
Date: Sun, 08 Jun 2014 16:22:18 +0300
Subject: [pypy-dev] PyPy 2.3.1 released
In-Reply-To: <CAMSv6X2=Hug6+Xg82irYP5gj2Z80RPNJa1=vt+9j+uQTRLq80A@mail.gmail.com>
References: <CAMSv6X2=Hug6+Xg82irYP5gj2Z80RPNJa1=vt+9j+uQTRLq80A@mail.gmail.com>
Message-ID: <1794958.6Ysl0DVEMg@dragon.dg>

Hi,

The links to the source tarballs are broken (on the download page).  

Regards

On Sunday, 8 June 2014 09:08:41 Armin Rigo wrote:
> Hi all,
> 
> We're pleased to announce PyPy 2.3.1, a feature-and-bugfix improvement
> over our recent 2.3 release last month.
> 
> 
> This release contains several bugfixes and enhancements among the
> user-facing improvements:
> 
> * The built-in struct module was renamed to _struct, solving issues
> with IDLE and other modules
> 
> * Support for compilation with gcc-4.9
> 
> * A CFFI-based version of the gdbm module is now included in our binary
> bundle
> 
> * Many issues were resolved since the 2.3 release on May 8
> 
> 
> Complete release notice: http://doc.pypy.org/en/latest/release-2.3.1.html
> Download: http://pypy.org/download.html
> 
> Please try it out and let us know what you think. We especially
> welcome success stories, please tell us about how it has helped you!
> 
> Cheers, The PyPy Team (thanks Matti!)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 964 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140608/aa6b5dc5/attachment.sig>

From matti.picus at gmail.com  Sun Jun  8 17:01:32 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Sun, 08 Jun 2014 18:01:32 +0300
Subject: [pypy-dev] PyPy 2.3.1 released
In-Reply-To: <1794958.6Ysl0DVEMg@dragon.dg>
References: <CAMSv6X2=Hug6+Xg82irYP5gj2Z80RPNJa1=vt+9j+uQTRLq80A@mail.gmail.com>
 <1794958.6Ysl0DVEMg@dragon.dg>
Message-ID: <53947ACC.10204@gmail.com>

Thanks, fixed. It may take a few minutes to percolate out, and then you 
need to refresh the cached page.
Matti

On 06/08/2014 04:22 PM, David Naylor wrote:
> Hi,
>
> The links to the source tarballs are broken (on the download page).
>
> Regards
>
> On Sunday, 8 June 2014 09:08:41 Armin Rigo wrote:
>> Hi all,
>>
>> We're pleased to announce PyPy 2.3.1, a feature-and-bugfix improvement
>> over our recent 2.3 release last month.
>>
>>
>> This release contains several bugfixes and enhancements among the
>> user-facing improvements:
>>
>> * The built-in struct module was renamed to _struct, solving issues
>> with IDLE and other modules
>>
>> * Support for compilation with gcc-4.9
>>
>> * A CFFI-based version of the gdbm module is now included in our binary
>> bundle
>>
>> * Many issues were resolved since the 2.3 release on May 8
>>
>>
>> Complete release notice: http://doc.pypy.org/en/latest/release-2.3.1.html
>> Download: http://pypy.org/download.html
>>
>> Please try it out and let us know what you think. We especially
>> welcome success stories, please tell us about how it has helped you!
>>
>> Cheers, The PyPy Team (thanks Matti!)
>>
>>
>> _______________________________________________
>> pypy-dev mailing list
>> pypy-dev at python.org
>> https://mail.python.org/mailman/listinfo/pypy-dev


From wpf710 at gmail.com  Mon Jun  9 07:18:23 2014
From: wpf710 at gmail.com (wpf)
Date: Mon, 9 Jun 2014 13:18:23 +0800
Subject: [pypy-dev] No Module _bsddb
Message-ID: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>

Hi All

  I am trying to use the Pypy under Ubuntu 12.04, and pypy version as below

~$ pypy --version
Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
[PyPy 2.3.0 with GCC 4.6.3]

-$ python --version
Python 2.7.3

and when I  try to import _bsddb in pypy, it complains that 'No module
named _bsddb'

but it works for python.

so how to resolve it? I 'd tried to google it ,but nothing useful found.

THanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140609/fb8ff398/attachment.html>

From benjamin at python.org  Mon Jun  9 07:29:38 2014
From: benjamin at python.org (Benjamin Peterson)
Date: Sun, 08 Jun 2014 22:29:38 -0700
Subject: [pypy-dev] No Module _bsddb
In-Reply-To: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
References: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
Message-ID: <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>

On Sun, Jun 8, 2014, at 22:18, wpf wrote:
> Hi All
> 
>   I am trying to use the Pypy under Ubuntu 12.04, and pypy version as
>   below
> 
> ~$ pypy --version
> Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
> [PyPy 2.3.0 with GCC 4.6.3]
> 
> -$ python --version
> Python 2.7.3
> 
> and when I  try to import _bsddb in pypy, it complains that 'No module
> named _bsddb'
> 
> but it works for python.
> 
> so how to resolve it? I 'd tried to google it ,but nothing useful found.

PyPy doesn't support the bsddb module.

From matti.picus at gmail.com  Mon Jun  9 20:41:17 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Mon, 09 Jun 2014 21:41:17 +0300
Subject: [pypy-dev] No Module _bsddb
In-Reply-To: <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
References: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
 <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
Message-ID: <5395FFCD.9070108@gmail.com>

It would seem to be pretty easy to support in the same way as we support 
gdbm: as a thin cffi wrapper around the Berkely DB shared object. This 
might be something a new developer could easily contribute...
Matti

On 06/09/2014 08:29 AM, Benjamin Peterson wrote:
> On Sun, Jun 8, 2014, at 22:18, wpf wrote:
>> Hi All
>>
>>    I am trying to use the Pypy under Ubuntu 12.04, and pypy version as
>>    below
>>
>> ~$ pypy --version
>> Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
>> [PyPy 2.3.0 with GCC 4.6.3]
>>
>> -$ python --version
>> Python 2.7.3
>>
>> and when I  try to import _bsddb in pypy, it complains that 'No module
>> named _bsddb'
>>
>> but it works for python.
>>
>> so how to resolve it? I 'd tried to google it ,but nothing useful found.
> PyPy doesn't support the bsddb module.
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev


From wpf710 at gmail.com  Tue Jun 10 04:14:55 2014
From: wpf710 at gmail.com (wpf)
Date: Tue, 10 Jun 2014 10:14:55 +0800
Subject: [pypy-dev] No Module _bsddb
In-Reply-To: <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
References: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
 <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
Message-ID: <CAF_H8kAqSEB=sDkh4WiVKZ76JDsQLjDnaixEbUEC27OfVMCamA@mail.gmail.com>

Hi Benjamin

    I understand that.

    anyway, when I checking the code as below

norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ which
pypy
/usr/bin/pypy
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls -lt
/usr/bin/pypy
lrwxrwxrwx 1 root root 22 May 20 19:57 /usr/bin/pypy ->
../lib/pypy/bin/pypy-c
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ pwd
/usr/lib/pypy/lib-python/2.7/bsddb
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls -lt
__init__.py
-rw-r--r-- 1 root root 15988 May 20 10:48 __init__.py

from the code of __init__.py , it's still trying to load _bsddb module,
 you know ,I am a newbie to python, did I misunderstand something?

and also I will try to understand why my code(actually ,it's openstack tox
env) will trigger that.





On Mon, Jun 9, 2014 at 1:29 PM, Benjamin Peterson <benjamin at python.org>
wrote:

> On Sun, Jun 8, 2014, at 22:18, wpf wrote:
> > Hi All
> >
> >   I am trying to use the Pypy under Ubuntu 12.04, and pypy version as
> >   below
> >
> > ~$ pypy --version
> > Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
> > [PyPy 2.3.0 with GCC 4.6.3]
> >
> > -$ python --version
> > Python 2.7.3
> >
> > and when I  try to import _bsddb in pypy, it complains that 'No module
> > named _bsddb'
> >
> > but it works for python.
> >
> > so how to resolve it? I 'd tried to google it ,but nothing useful found.
>
> PyPy doesn't support the bsddb module.
>



-- 
??!

   ???
    msn: dragoninfield at hotmail.com
    mobile: 13681265240
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140610/95722e64/attachment.html>

From radix at twistedmatrix.com  Tue Jun 10 05:08:46 2014
From: radix at twistedmatrix.com (Christopher Armstrong)
Date: Mon, 9 Jun 2014 22:08:46 -0500
Subject: [pypy-dev] No Module _bsddb
In-Reply-To: <CAF_H8kAqSEB=sDkh4WiVKZ76JDsQLjDnaixEbUEC27OfVMCamA@mail.gmail.com>
References: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
 <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
 <CAF_H8kAqSEB=sDkh4WiVKZ76JDsQLjDnaixEbUEC27OfVMCamA@mail.gmail.com>
Message-ID: <etPan.539676bf.643c9869.6143@densetsu>

On June 9, 2014 at 9:15:30 PM, wpf (wpf710 at gmail.com) wrote:
Hi Benjamin

? ? I understand that.?

? ? anyway, when I checking the code as below

norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ which pypy
/usr/bin/pypy
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls -lt /usr/bin/pypy
lrwxrwxrwx 1 root root 22 May 20 19:57 /usr/bin/pypy -> ../lib/pypy/bin/pypy-c
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ pwd
/usr/lib/pypy/lib-python/2.7/bsddb
norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls -lt __init__.py
-rw-r--r-- 1 root root 15988 May 20 10:48 __init__.py

from the code of __init__.py , it's still trying to load _bsddb module, ?you know ,I am a newbie to python, did I misunderstand something?

and also I will try to understand why my code(actually ,it's openstack tox env) will trigger that.




This may be happening because you?ve run tests with normal python and then switched to pypy. Some of the testing tools used by openstack create a .testrepository directory with data files based on bsddb if it?s available, but use something else if it?s not. Try deleting your .testrepository directory and trying again.


--?
Christopher Armstrong
http://twitter.com/radix
http://wordeology.com/




On Mon, Jun 9, 2014 at 1:29 PM, Benjamin Peterson <benjamin at python.org> wrote:
On Sun, Jun 8, 2014, at 22:18, wpf wrote:
> Hi All
>
> ? I am trying to use the Pypy under Ubuntu 12.04, and pypy version as
> ? below
>
> ~$ pypy --version
> Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
> [PyPy 2.3.0 with GCC 4.6.3]
>
> -$ python --version
> Python 2.7.3
>
> and when I ?try to import _bsddb in pypy, it complains that 'No module
> named _bsddb'
>
> but it works for python.
>
> so how to resolve it? I 'd tried to google it ,but nothing useful found.

PyPy doesn't support the bsddb module.



--
??!
?
?? ???
? ? msn: dragoninfield at hotmail.com
? ? mobile: 13681265240
_______________________________________________  
pypy-dev mailing list  
pypy-dev at python.org  
https://mail.python.org/mailman/listinfo/pypy-dev  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140609/2b456b0a/attachment.html>

From wpf710 at gmail.com  Tue Jun 10 05:31:36 2014
From: wpf710 at gmail.com (wpf)
Date: Tue, 10 Jun 2014 11:31:36 +0800
Subject: [pypy-dev] No Module _bsddb
In-Reply-To: <etPan.539676bf.643c9869.6143@densetsu>
References: <CAF_H8kBQ5VPNDw-dcE6tNOZscP2oWaa8MESv+4mLmwwg404JuQ@mail.gmail.com>
 <1402291778.962.126610293.1B1D79B3@webmail.messagingengine.com>
 <CAF_H8kAqSEB=sDkh4WiVKZ76JDsQLjDnaixEbUEC27OfVMCamA@mail.gmail.com>
 <etPan.539676bf.643c9869.6143@densetsu>
Message-ID: <CAF_H8kAdM78K1BfWOJBf-ejBMLGNxCy7s3o9RLadbx+r9ZjxZg@mail.gmail.com>

That really help!!

Thanks a lot.


On Tue, Jun 10, 2014 at 11:08 AM, Christopher Armstrong <
radix at twistedmatrix.com> wrote:

> On June 9, 2014 at 9:15:30 PM, wpf (wpf710 at gmail.com) wrote:
>
> Hi Benjamin
>
>     I understand that.
>
>     anyway, when I checking the code as below
>
>  norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ which
> pypy
> /usr/bin/pypy
>  norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls
> -lt /usr/bin/pypy
> lrwxrwxrwx 1 root root 22 May 20 19:57 /usr/bin/pypy ->
> ../lib/pypy/bin/pypy-c
>  norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ pwd
> /usr/lib/pypy/lib-python/2.7/bsddb
>  norman at norman-ThinkCentre-M92P:/usr/lib/pypy/lib-python/2.7/bsddb$ ls
> -lt __init__.py
>  -rw-r--r-- 1 root root 15988 May 20 10:48 __init__.py
>
> from the code of __init__.py , it's still trying to load _bsddb module,
>  you know ,I am a newbie to python, did I misunderstand something?
>
> and also I will try to understand why my code(actually ,it's openstack tox
> env) will trigger that.
>
>
>
>
> This may be happening because you?ve run tests with normal python and then
> switched to pypy. Some of the testing tools used by openstack create a
> .testrepository directory with data files based on bsddb if it?s available,
> but use something else if it?s not. Try deleting your .testrepository
> directory and trying again.
>
>
> --
> Christopher Armstrong
> http://twitter.com/radix
> http://wordeology.com/
>
>
>
>
> On Mon, Jun 9, 2014 at 1:29 PM, Benjamin Peterson <benjamin at python.org>
> wrote:
>
>>  On Sun, Jun 8, 2014, at 22:18, wpf wrote:
>> > Hi All
>> >
>> >   I am trying to use the Pypy under Ubuntu 12.04, and pypy version as
>> >   below
>> >
>> > ~$ pypy --version
>> > Python 2.7.6 (2.3+dfsg-1~ppa0, May 20 2014, 09:11:18)
>> > [PyPy 2.3.0 with GCC 4.6.3]
>> >
>> > -$ python --version
>> > Python 2.7.3
>> >
>> > and when I  try to import _bsddb in pypy, it complains that 'No module
>> > named _bsddb'
>> >
>> > but it works for python.
>> >
>> > so how to resolve it? I 'd tried to google it ,but nothing useful found.
>>
>>  PyPy doesn't support the bsddb module.
>>
>
>
>
> --
> ??!
>
>    ???
>     msn: dragoninfield at hotmail.com
>     mobile: 13681265240
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
>
>


-- 
??!

   ???
    msn: dragoninfield at hotmail.com
    mobile: 13681265240
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140610/2d0b8e3a/attachment-0001.html>

From usethisid2014 at gmail.com  Wed Jun 11 07:32:31 2014
From: usethisid2014 at gmail.com (Gayathri J)
Date: Wed, 11 Jun 2014 11:02:31 +0530
Subject: [pypy-dev] Numpy on pypy
Message-ID: <CAKiaUYyHmaMjpF4OY4o=+pexKt5kaXkuEFqUvFm+Ofwufx8VxQ@mail.gmail.com>

Hi

I am trying to do a matrix dot using numpy.dot

the numpy.dot on pypy seems lot slower than bumpy.do on python
is there an alternative to using numpy.dot - so that i can do matrix.dot on
pypy faster than cpython....

suggestions would be very helpful

thanks Gayathri
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140611/265b1606/attachment.html>

From usethisid2014 at gmail.com  Wed Jun 11 18:07:17 2014
From: usethisid2014 at gmail.com (Gayathri J)
Date: Wed, 11 Jun 2014 21:37:17 +0530
Subject: [pypy-dev] Matrix dot product
Message-ID: <CAKiaUYyd2j8-3EMGi37SNP_PKJKLQ=MtW4x1NVWskWfr746v5A@mail.gmail.com>

Following is my python code for matrix multiplication


*import time*
*a=random.rand(1**28,128,128,15)*
*b=np.random.rand(15,15)*
*c=np.zeros((128,128,128,15), dtype=np.float64)*
*t=time.time()*
*c=a.dot(b)*
*print time.time()-t*


Timing for

*pypy -> 3.957*
*python 2.7 -> .80*

So find that the basic matrix dot operation is faster on python 2.7 than on
pypy 2.2.1
why is this so? am i missing something?

thanks gayathri
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140611/3b512e37/attachment.html>

From skip at pobox.com  Wed Jun 11 19:12:36 2014
From: skip at pobox.com (Skip Montanaro)
Date: Wed, 11 Jun 2014 12:12:36 -0500
Subject: [pypy-dev] Matrix dot product
In-Reply-To: <CAKiaUYyd2j8-3EMGi37SNP_PKJKLQ=MtW4x1NVWskWfr746v5A@mail.gmail.com>
References: <CAKiaUYyd2j8-3EMGi37SNP_PKJKLQ=MtW4x1NVWskWfr746v5A@mail.gmail.com>
Message-ID: <CANc-5UyRU=n4LYF0nkLfLb4HfDjAq2WrBfTjAhBM2zvjfuu9yQ@mail.gmail.com>

On Wed, Jun 11, 2014 at 11:07 AM, Gayathri J <usethisid2014 at gmail.com> wrote:
> So find that the basic matrix dot operation is faster on python 2.7 than on
> pypy 2.2.1
> why is this so? am i missing something?

Warning: I am not a PyPy user, just an observer.

It looks to me like you haven't executed the dot product code enough
for PyPy's tracing code to recognize it as a critical path.
Consequently, you have only suffered from the JIT machinery startup
and execution tracing overhead, and haven't benefited from the JIT
compilation. You might try wrapping a loop around the a.dot(b) call:

for i in range(1000):
    c = a.dot(b)

I have a couple more nits which are not PyPy-specific:

1. When asking for help, it really helps to post code that actually
runs. Your code doesn't include the numpy import. You also must of
left off the "np." prefix in the assignment to a.

2. You don't need the c = np.zeros(...) call either. You just throw it
away two lines later.

Skip

From matti.picus at gmail.com  Wed Jun 11 19:10:26 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Wed, 11 Jun 2014 20:10:26 +0300
Subject: [pypy-dev] Matrix dot product
In-Reply-To: <CAKiaUYyd2j8-3EMGi37SNP_PKJKLQ=MtW4x1NVWskWfr746v5A@mail.gmail.com>
References: <CAKiaUYyd2j8-3EMGi37SNP_PKJKLQ=MtW4x1NVWskWfr746v5A@mail.gmail.com>
Message-ID: <53988D82.2030009@gmail.com>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140611/66fd33bd/attachment.html>

From john.barry at pervasive-health.com  Wed Jun 11 23:27:04 2014
From: john.barry at pervasive-health.com (Barry, John)
Date: Wed, 11 Jun 2014 16:27:04 -0500
Subject: [pypy-dev] --sandbox woes. H ow do I install modules?
Message-ID: <CAJhU4gA72m-RO7aZeJY9iAo2t8ZLFNQ-VtTdRm_08Enm4QhCRA@mail.gmail.com>

Hoping someone here can point me in the right direction.


I'm attempting to create a sandbox, load it up with our requirements via
pip, and start figuring out how to execute this from our codebase.

It all falls down every single time when I attempt to install
distribute.py, because import time fails within the sandbox.

So the question is twofold:
(1) How can I get import time working (as we call
datetime.datetime.utcnow() with astonishing regularity)
(2) How do I install additional packages into the sandbox if I can't
install pip (potentially not an issue if import time reoslves it)

-- 


Confidentiality Notice: This e-mail transmission may contain confidential 
or legally privileged information that is intended only for the individual 
or entity named in the e-mail address. If you have received this 
communication in error, please notify me by return e-mail, and destroy this 
communication and all copies thereof, including any attachments. 

Apervita ? is a registered trademark of Pervasive Health Inc.

 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140611/0daec144/attachment.html>

From arigo at tunes.org  Thu Jun 12 11:00:00 2014
From: arigo at tunes.org (Armin Rigo)
Date: Thu, 12 Jun 2014 11:00:00 +0200
Subject: [pypy-dev] Numpy on pypy
In-Reply-To: <CAKiaUYyHmaMjpF4OY4o=+pexKt5kaXkuEFqUvFm+Ofwufx8VxQ@mail.gmail.com>
References: <CAKiaUYyHmaMjpF4OY4o=+pexKt5kaXkuEFqUvFm+Ofwufx8VxQ@mail.gmail.com>
Message-ID: <CAMSv6X0B-UNf9gkrywm-AV5gbu-huM12kGYDW-xrNLVG9e=oOw@mail.gmail.com>

Hi,

On 11 June 2014 07:32, Gayathri J <usethisid2014 at gmail.com> wrote:
> I am trying to do a matrix dot using numpy.dot
>
> the numpy.dot on pypy seems lot slower than bumpy.do on python
> is there an alternative to using numpy.dot - so that i can do matrix.dot on
> pypy faster than cpython....

For reference, I think you are misunderstanding PyPy somewhere.
Assume that there existed a general-purpose self-contained algorithm
to run matrix.dot(a,b) 10 times faster, written in C or Fortran.
Don't you think that the same algorithm would be used by CPython's
numpy too?

PyPy can't magically speed up a _single_ operation that takes a lot of
time.  What PyPy can do is optimize some complex Python code between
such operations.  In some cases it can also select more optimized
versions of the operations based on information from the Python
context.

For example, if you are timing how long it takes to do ``x =
3**10000000``, then there is no chance it would be faster on PyPy than
on CPython.  But if you're writing in Python an algorithm that
computes the same result, then this Python code would run much faster
on PyPy than on CPython.

Another example is that the time it takes to compare two very big
strings should be the same; but if we know that one string is always
"A", then the comparison ``s == "A"`` can be optimized by selecting a
special string comparison for length one.


A bient?t,

Armin.

From running.sun at gmail.com  Thu Jun 12 22:18:17 2014
From: running.sun at gmail.com (nejaka osoba)
Date: Thu, 12 Jun 2014 22:18:17 +0200
Subject: [pypy-dev] my notes about trying to build unipycation
Message-ID: <CAK29KF8VBectPczBuiuSdkSwmDVFA8GE0SkU=oms=aMeXxzYmw@mail.gmail.com>

https://gist.github.com/koo5/393746b19624debb2472

From matti.picus at gmail.com  Thu Jun 12 23:24:14 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Fri, 13 Jun 2014 00:24:14 +0300
Subject: [pypy-dev] failing test for on windows
Message-ID: <539A1A7E.3060705@gmail.com>

We are down to the last 50 own tests on win32.
Here is one I need help with:
http://buildbot.pypy.org/summary/longrepr?testname=TestLowLevelType.%28%29.test_round_up_for_allocation&builder=own-win-x86-32&build=162&mod=translator.c.test.test_lltyped

The failure is that, like ARM, ssize == llmemory.sizeof(Signed) * 2 
where ssize is 8
Is the test wrong or is there a problem with overallocating memory?
Matti

From matti.picus at gmail.com  Thu Jun 12 23:21:10 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Fri, 13 Jun 2014 00:21:10 +0300
Subject: [pypy-dev] test_round_up_for_allocation is failing on windows
Message-ID: <539A19C6.20004@gmail.com>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140613/3bc4c900/attachment.html>

From matti.picus at gmail.com  Fri Jun 13 15:10:30 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Fri, 13 Jun 2014 16:10:30 +0300
Subject: [pypy-dev] PYPYLOG and colon as a special character
Message-ID: <539AF846.2020603@gmail.com>

if PYPYLOG is a filename, we log to it (except debug_print)
if PYPYLOG is :filename, we log debug_print as well
if PYPYLOG is section:filename, we log only that section

But : is a valid char for filename on windows, so something like
PYPYLOG=C:\TEMP\log.log
does the wrong thing, and even worse, does so silently.

Matti

From chrish at pianocktail.org  Fri Jun 13 16:21:39 2014
From: chrish at pianocktail.org (Christian Hudon)
Date: Fri, 13 Jun 2014 10:21:39 -0400
Subject: [pypy-dev] PYPYLOG and colon as a special character
In-Reply-To: <539AF846.2020603@gmail.com>
References: <539AF846.2020603@gmail.com>
Message-ID: <0c88c477fded5d1448af6854096882f5@chrish.pianocktail.org>

The colon character is valid in paths in Windows (because of the drive 
letter), but not in filenames (it's used for supporting different 
streams, which hardly anyone uses). So doing that check on 
os.path.basename(os.getenv("PYPYLOG")) should work under Windows also.

Of course, I didn't check the code, so it's possible I'm missing 
something too.

On 2014-06-13 9:10, Matti Picus wrote:
> if PYPYLOG is a filename, we log to it (except debug_print)
> if PYPYLOG is :filename, we log debug_print as well
> if PYPYLOG is section:filename, we log only that section
>
> But : is a valid char for filename on windows, so something like
> PYPYLOG=C:\TEMP\log.log
> does the wrong thing, and even worse, does so silently.
>
> Matti
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev


From matti.picus at gmail.com  Fri Jun 13 16:34:13 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Fri, 13 Jun 2014 17:34:13 +0300
Subject: [pypy-dev] PYPYLOG and colon as a special character
In-Reply-To: <0c88c477fded5d1448af6854096882f5@chrish.pianocktail.org>
References: <539AF846.2020603@gmail.com>
 <0c88c477fded5d1448af6854096882f5@chrish.pianocktail.org>
Message-ID: <539B0BE5.3070207@gmail.com>

We discussed this on IRC and decided to add a workaround, '+filename' 
will be handled like 'filename'
Matti

On 13/06/2014 5:21 PM, Christian Hudon wrote:
> The colon character is valid in paths in Windows (because of the drive 
> letter), but not in filenames (it's used for supporting different 
> streams, which hardly anyone uses). So doing that check on 
> os.path.basename(os.getenv("PYPYLOG")) should work under Windows also.
>
> Of course, I didn't check the code, so it's possible I'm missing 
> something too.
>
> On 2014-06-13 9:10, Matti Picus wrote:
>> if PYPYLOG is a filename, we log to it (except debug_print)
>> if PYPYLOG is :filename, we log debug_print as well
>> if PYPYLOG is section:filename, we log only that section
>>
>> But : is a valid char for filename on windows, so something like
>> PYPYLOG=C:\TEMP\log.log
>> does the wrong thing, and even worse, does so silently.
>>
>> Matti
>> _______________________________________________
>> pypy-dev mailing list
>> pypy-dev at python.org
>> https://mail.python.org/mailman/listinfo/pypy-dev
>
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev


From arigo at tunes.org  Fri Jun 13 17:08:07 2014
From: arigo at tunes.org (Armin Rigo)
Date: Fri, 13 Jun 2014 17:08:07 +0200
Subject: [pypy-dev] PYPYLOG and colon as a special character
In-Reply-To: <0c88c477fded5d1448af6854096882f5@chrish.pianocktail.org>
References: <539AF846.2020603@gmail.com>
 <0c88c477fded5d1448af6854096882f5@chrish.pianocktail.org>
Message-ID: <CAMSv6X2HbO2bnxbDBv+SryfsKSntdcZO1V8wwXhT+Vdn1ukdEw@mail.gmail.com>

Hi Christian,

On 13 June 2014 16:21, Christian Hudon <chrish at pianocktail.org> wrote:
> The colon character is valid in paths in Windows (because of the drive
> letter), but not in filenames (it's used for supporting different streams,
> which hardly anyone uses). So doing that check on
> os.path.basename(os.getenv("PYPYLOG")) should work under Windows also.

You're missing the fact that "C:\Foo" might represent either a path
name (case 1), or (case 3) "section:filename", with section "C" and
path "\Foo".


A bient?t,

Armin.

From arigo at tunes.org  Sat Jun 14 18:26:30 2014
From: arigo at tunes.org (Armin Rigo)
Date: Sat, 14 Jun 2014 18:26:30 +0200
Subject: [pypy-dev] --sandbox woes. H ow do I install modules?
In-Reply-To: <CAJhU4gA72m-RO7aZeJY9iAo2t8ZLFNQ-VtTdRm_08Enm4QhCRA@mail.gmail.com>
References: <CAJhU4gA72m-RO7aZeJY9iAo2t8ZLFNQ-VtTdRm_08Enm4QhCRA@mail.gmail.com>
Message-ID: <CAMSv6X0nQZJ+D+2eiJXpQKUDrphT7D5WyPDdLRY2k8VuKy-Jhw@mail.gmail.com>

Hi John,

On 11 June 2014 23:27, Barry, John <john.barry at pervasive-health.com> wrote:
> I'm attempting to create a sandbox, load it up with our requirements via
> pip, and start figuring out how to execute this from our codebase.
>
> It all falls down every single time when I attempt to install distribute.py,
> because import time fails within the sandbox.

That's expected: I would not expect pip to work *at all* in a
sandboxed PyPy.  Now more precisely, the "time" module should be
present in a reduced format, containing only three functions; you seem
to say that it is completely absent, which I find strange but
certainly not impossible as we don't test the sandbox regularly.

But speaking more generally, most standard built-in modules are absent
(or very reduced, like "time") for now.  The idea is that you run
small snippets of code provided by some untrusted user, doing some
computation in Python.  Ideally you don't need a lot of performance,
so you can leave out the JIT, whose presence could in theory introduce
additional security bugs.

Now if someone really needs to run a large
program with many dependencies, then he needs to invest some more
serious amount of time into pypy's sandbox, and finish it at least
until all modules that he really needs are supported.  This shouldn't
be too hard -- and, more importantly, it should be easy to make it
work *in a secure way* -- but it is time-consuming.  We haven't found
anyone interested in doing this so far.


A bient?t,

Armin.

From matti at picus.org.il  Sat Jun 14 23:54:44 2014
From: matti at picus.org.il (Matti Picus)
Date: Sun, 15 Jun 2014 00:54:44 +0300
Subject: [pypy-dev] wierd test failure when running tests in parallel
Message-ID: <539CC4A4.8030606@picus.org.il>

on windows own build, tests that call ctypes2lltype with an Unsigned 
failed last night's build, for instance
http://buildbot.pypy.org/summary/longrepr?testname=test_connect_ex&builder=own-win-x86-32&build=164&mod=rlib.test.test_rsocket
Running this test alone passes.
Two questions
1. Should there be a fastpath like
elif T is lltype.Unsigned:
     llobj = cobj
2. how can ctypes2lltype(Unsigned, 784L) fail?

Note: I have started running two test processes in parallel, which may 
be creating many more open file handles.

Matti

From arigo at tunes.org  Sun Jun 15 10:06:48 2014
From: arigo at tunes.org (Armin Rigo)
Date: Sun, 15 Jun 2014 10:06:48 +0200
Subject: [pypy-dev] wierd test failure when running tests in parallel
In-Reply-To: <539CC4A4.8030606@picus.org.il>
References: <539CC4A4.8030606@picus.org.il>
Message-ID: <CAMSv6X31aOQOsPx_8H9f5YVtSrjC2A3dz+Z9Zta3h7-cXX+KYA@mail.gmail.com>

Hi Matti,

On 14 June 2014 23:54, Matti Picus <matti at picus.org.il> wrote:
> 2. how can ctypes2lltype(Unsigned, 784L) fail?

Shooting in the dark, maybe running multiple tests in parallel caused
interference which meant that 'rffi.platform.numbertype_to_rclass' is
incompletely populated?


Armin

From arigo at tunes.org  Mon Jun 16 14:29:02 2014
From: arigo at tunes.org (Armin Rigo)
Date: Mon, 16 Jun 2014 14:29:02 +0200
Subject: [pypy-dev] pypy-stm: not dead
Message-ID: <CAMSv6X3YoasyMq07cSrhcyu5bjtPOROLj_g_fizbgL1ORSMWSQ@mail.gmail.com>

Hi all,

Just a quick note to tell you sorry: although the 2.3 release of PyPy
said the release of PyPy-STM would occur very soon, it's still
delayed.  We have found some real use cases in which it is several
times slower than PyPy (both with the JIT) and so we're busy setting
up a few missing things with the GC, notably the card marking scheme,
which we think is causing it.


Armin

From arigo at tunes.org  Mon Jun 16 14:33:40 2014
From: arigo at tunes.org (Armin Rigo)
Date: Mon, 16 Jun 2014 14:33:40 +0200
Subject: [pypy-dev] pypy-stm: not dead
In-Reply-To: <CAMSv6X3YoasyMq07cSrhcyu5bjtPOROLj_g_fizbgL1ORSMWSQ@mail.gmail.com>
References: <CAMSv6X3YoasyMq07cSrhcyu5bjtPOROLj_g_fizbgL1ORSMWSQ@mail.gmail.com>
Message-ID: <CAMSv6X2VrOJGz+eE6RfmrOnaNK61n=jhb9H86-NVqGO3A_KjBA@mail.gmail.com>

Hi again,

We have also been "delayed" by writing two papers about STM :-)
Thanks to Remi for doing the main work.  The first is a position paper
already accepted at ICOOOLPS'14:

http://bitbucket.org/pypy/extradoc/raw/extradoc/talk/icooolps2014/position-paper.pdf

The second one is submitted for DLS'14:

http://bitbucket.org/pypy/extradoc/raw/extradoc/talk/dls2014/paper/paper.pdf


A bient?t,

Armin.

From matti.picus at gmail.com  Tue Jun 17 22:01:20 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Tue, 17 Jun 2014 23:01:20 +0300
Subject: [pypy-dev] how can a class method become frozen in tests only on
	windows?
Message-ID: <53A09E90.5080502@gmail.com>

in looking at the own test_rzipfile failures on windows[0], the 
annotator is failing in rtyper.rpbc.redispatch_call since the method 
being called is a MultipleFrozenPBCRepr .
The method is flush_buffers from (it seems)  streamio.Stream [1]

Any hints how this can happen?
Matti


[0] 
http://buildbot.pypy.org/summary/longrepr?testname=TestRZipFile.%28%29.test_rzipfile&builder=own-win-x86-32&build=167&mod=rlib.test.test_rzipfile
[1] I added --pdb to the test run, and here is what the debugger had to say:
 > d:\pypy_stuff\pypy\rpython\rtyper\rpbc.py(825)redispatch_call()
-> SmallFunctionSetPBCRepr))
(Pdb) p self
<MethodsPBCRepr * GcStruct rpython.rlib.streamio.Stream { super }>
(Pdb) p self.methodname
'flush_buffers'

From cclauss at me.com  Wed Jun 18 03:19:43 2014
From: cclauss at me.com (cclauss)
Date: Wed, 18 Jun 2014 03:19:43 +0200
Subject: [pypy-dev] Support for current versions of Pypy on Heroku and Cloud
 Foundry platforms?
Message-ID: <F2E9CBAE-8567-4B2B-A327-869345B9A1D5@me.com>

Hi Folks,

    Currently Heroku only has Pypy 1.9 on an unsupported, experimental basis.

https://devcenter.heroku.com/articles/python-runtimes#supported-python-runtimes
and https://github.com/heroku/heroku-buildpack-python/issues/139

    Other Cloud Foundry-based IaaS offering (IBM BlueMix, etc.) seem to use similar buildpacks.
https://github.com/cloudfoundry-community/cf-docs-contrib/wiki/Buildpacks

    Do any of you know of an Infrastructure as a Service provider that supports current versions of Pypy?

Thanks for any pointers that you can provide.  Chris Clauss

From alex.gaynor at gmail.com  Wed Jun 18 07:06:40 2014
From: alex.gaynor at gmail.com (Alex Gaynor)
Date: Tue, 17 Jun 2014 22:06:40 -0700
Subject: [pypy-dev] Support for current versions of Pypy on Heroku and
 Cloud Foundry platforms?
In-Reply-To: <F2E9CBAE-8567-4B2B-A327-869345B9A1D5@me.com>
References: <F2E9CBAE-8567-4B2B-A327-869345B9A1D5@me.com>
Message-ID: <CAFRnB2Ub-7gO=__WNhUNk=qVDMUbYPhqxtQzVSaUwjR5yiv1gQ@mail.gmail.com>

Hi Chris,

Are you looking for an Infrastructure as a Service (something like AWS, or
Rackspace Cloud) or a Platform as a Service (Heroku)?

Typically IaaS providers just give you a bare linux box, where you can of
course install your own PyPy; as you've seen it looks like the PyPy on
Heroku is out of date. I think the easiest move is probably to look into
writing a custom build pack for Heroku, if that's what you're interested in.

Alex

PS: Disclaimer, I work at Rackspace.


On Tue, Jun 17, 2014 at 6:19 PM, cclauss <cclauss at me.com> wrote:

> Hi Folks,
>
>     Currently Heroku only has Pypy 1.9 on an unsupported, experimental
> basis.
>
>
> https://devcenter.heroku.com/articles/python-runtimes#supported-python-runtimes
> and https://github.com/heroku/heroku-buildpack-python/issues/139
>
>     Other Cloud Foundry-based IaaS offering (IBM BlueMix, etc.) seem to
> use similar buildpacks.
> https://github.com/cloudfoundry-community/cf-docs-contrib/wiki/Buildpacks
>
>     Do any of you know of an Infrastructure as a Service provider that
> supports current versions of Pypy?
>
> Thanks for any pointers that you can provide.  Chris Clauss
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
>



-- 
"I disapprove of what you say, but I will defend to the death your right to
say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
"The people's good is the highest law." -- Cicero
GPG Key fingerprint: 125F 5C67 DFE9 4084
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140617/145a1a0a/attachment.html>

From cclauss at me.com  Wed Jun 18 08:01:53 2014
From: cclauss at me.com (cclauss)
Date: Wed, 18 Jun 2014 08:01:53 +0200
Subject: [pypy-dev] Support for current versions of Pypy on Heroku and
 Cloud Foundry platforms?
In-Reply-To: <CAFRnB2Ub-7gO=__WNhUNk=qVDMUbYPhqxtQzVSaUwjR5yiv1gQ@mail.gmail.com>
References: <F2E9CBAE-8567-4B2B-A327-869345B9A1D5@me.com>
 <CAFRnB2Ub-7gO=__WNhUNk=qVDMUbYPhqxtQzVSaUwjR5yiv1gQ@mail.gmail.com>
Message-ID: <29183C4F-26D9-43B1-BDB2-3A067992F80B@me.com>

PaaS is what I am looking for.

On 18 Jun 2014, at 07:06, Alex Gaynor <alex.gaynor at gmail.com> wrote:

> Hi Chris,
> 
> Are you looking for an Infrastructure as a Service (something like AWS, or Rackspace Cloud) or a Platform as a Service (Heroku)?
> 
> Typically IaaS providers just give you a bare linux box, where you can of course install your own PyPy; as you've seen it looks like the PyPy on Heroku is out of date. I think the easiest move is probably to look into writing a custom build pack for Heroku, if that's what you're interested in.
> 
> Alex
> 
> PS: Disclaimer, I work at Rackspace.
> 
> 
> On Tue, Jun 17, 2014 at 6:19 PM, cclauss <cclauss at me.com> wrote:
> Hi Folks,
> 
>     Currently Heroku only has Pypy 1.9 on an unsupported, experimental basis.
> 
> https://devcenter.heroku.com/articles/python-runtimes#supported-python-runtimes
> and https://github.com/heroku/heroku-buildpack-python/issues/139
> 
>     Other Cloud Foundry-based IaaS offering (IBM BlueMix, etc.) seem to use similar buildpacks.
> https://github.com/cloudfoundry-community/cf-docs-contrib/wiki/Buildpacks
> 
>     Do any of you know of an Infrastructure as a Service provider that supports current versions of Pypy?
> 
> Thanks for any pointers that you can provide.  Chris Clauss
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
> 
> 
> 
> -- 
> "I disapprove of what you say, but I will defend to the death your right to say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
> "The people's good is the highest law." -- Cicero
> GPG Key fingerprint: 125F 5C67 DFE9 4084

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140618/6af1d6df/attachment.html>

From mail at jdborg.com  Thu Jun 19 10:28:26 2014
From: mail at jdborg.com (=?UTF-8?B?Sm9lIEJvcsSh?=)
Date: Thu, 19 Jun 2014 09:28:26 +0100
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
Message-ID: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>

Hi all,

Because I've had to build zlib away from its normal home, I need to pass -I
to gcc (LD_LIBRARY_PATH doesn't work).

Setting this -I fixes my build issue:

gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
/tmp/usession-release-2.3.1-3/platcheck_57.c -o
/tmp/usession-release-2.3.1-3/platcheck_57.o
/tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h:
No such file or director

gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
-I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
/tmp/usession-release-2.3.1-3/platcheck_57.o


Thanks,
Joseph David Bor?
josephb.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140619/0a73f863/attachment.html>

From rymg19 at gmail.com  Thu Jun 19 18:05:00 2014
From: rymg19 at gmail.com (Ryan Gonzalez)
Date: Thu, 19 Jun 2014 11:05:00 -0500
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
In-Reply-To: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
References: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
Message-ID: <CAO41-mN17LwNkvuEmeQK-_j7RsPpkATX=TAMACr=HY4qooFWrQ@mail.gmail.com>

NOTE: I am not a PyPy developer.

Looking at RPython's translator(rpython/translator/c/genc.py), can't you
set the CFLAGSEXTRA environment variable with the extra flags?


On Thu, Jun 19, 2014 at 3:28 AM, Joe Bor? <mail at jdborg.com> wrote:

> Hi all,
>
> Because I've had to build zlib away from its normal home, I need to pass
> -I to gcc (LD_LIBRARY_PATH doesn't work).
>
> Setting this -I fixes my build issue:
>
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
> /tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h:
> No such file or director
>
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> -I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
>
>
> Thanks,
> Joseph David Bor?
> josephb.org
>
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
>
>


-- 
Ryan
If anybody ever asks me why I prefer C++ to C, my answer will be simple:
"It's becauseslejfp23(@#Q*(E*EIdc-SEGFAULT. Wait, I don't think that was
nul-terminated."
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140619/54dd6b94/attachment.html>

From romain.py at gmail.com  Thu Jun 19 19:48:25 2014
From: romain.py at gmail.com (Romain Guillebert)
Date: Thu, 19 Jun 2014 19:48:25 +0200
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
In-Reply-To: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
References: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
Message-ID: <20140619174825.GC20988@Plop>

Hi Joe

Isn't LIBRARY_PATH the proper variable for finding libraries at compile
time ?

Cheers
Romain

On 06/19, Joe Bor? wrote:
> Hi all,
> 
> Because I've had to build zlib away from its normal home, I need to pass -I
> to gcc (LD_LIBRARY_PATH doesn't work).
> 
> Setting this -I fixes my build issue:
> 
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
> /tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h:
> No such file or director
> 
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> -I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
> 
> 
> Thanks,
> Joseph David Bor?
> josephb.org

> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev


From yyc1992 at gmail.com  Fri Jun 20 00:29:38 2014
From: yyc1992 at gmail.com (Yichao Yu)
Date: Fri, 20 Jun 2014 06:29:38 +0800
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
In-Reply-To: <20140619174825.GC20988@Plop>
References: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
 <20140619174825.GC20988@Plop>
Message-ID: <CAMvDr+R8YK-VF3Cxuknk7m+6+AZXKtPuNrYKLzgP4X-a_Jwg7A@mail.gmail.com>

On Fri, Jun 20, 2014 at 1:48 AM, Romain Guillebert <romain.py at gmail.com> wrote:
> Hi Joe
>
> Isn't LIBRARY_PATH the proper variable for finding libraries at compile
> time ?

But isn't your problem finding the header files? (Thus -I.) ?

>
> Cheers
> Romain
>
> On 06/19, Joe Bor? wrote:
>> Hi all,
>>
>> Because I've had to build zlib away from its normal home, I need to pass -I
>> to gcc (LD_LIBRARY_PATH doesn't work).
>>
>> Setting this -I fixes my build issue:
>>
>> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
>> /tmp/usession-release-2.3.1-3/platcheck_57.c -o
>> /tmp/usession-release-2.3.1-3/platcheck_57.o
>> /tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h:
>> No such file or director
>>
>> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
>> -I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
>> /tmp/usession-release-2.3.1-3/platcheck_57.o
>>
>>
>> Thanks,
>> Joseph David Bor?
>> josephb.org
>
>> _______________________________________________
>> pypy-dev mailing list
>> pypy-dev at python.org
>> https://mail.python.org/mailman/listinfo/pypy-dev
>
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev

From romain.py at gmail.com  Fri Jun 20 00:54:03 2014
From: romain.py at gmail.com (Romain Guillebert)
Date: Fri, 20 Jun 2014 00:54:03 +0200
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
In-Reply-To: <CAMvDr+R8YK-VF3Cxuknk7m+6+AZXKtPuNrYKLzgP4X-a_Jwg7A@mail.gmail.com>
References: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
 <20140619174825.GC20988@Plop>
 <CAMvDr+R8YK-VF3Cxuknk7m+6+AZXKtPuNrYKLzgP4X-a_Jwg7A@mail.gmail.com>
Message-ID: <20140619225403.GD20988@Plop>

Yes, sorry then the variable is CPATH
(https://gcc.gnu.org/onlinedocs/gcc/Environment-Variables.html)

On 06/20, Yichao Yu wrote:
> On Fri, Jun 20, 2014 at 1:48 AM, Romain Guillebert <romain.py at gmail.com> wrote:
> > Hi Joe
> >
> > Isn't LIBRARY_PATH the proper variable for finding libraries at compile
> > time ?
> 
> But isn't your problem finding the header files? (Thus -I.) ?
> 
> >
> > Cheers
> > Romain
> >
> > On 06/19, Joe Bor? wrote:
> >> Hi all,
> >>
> >> Because I've had to build zlib away from its normal home, I need to pass -I
> >> to gcc (LD_LIBRARY_PATH doesn't work).
> >>
> >> Setting this -I fixes my build issue:
> >>
> >> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> >> /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> >> /tmp/usession-release-2.3.1-3/platcheck_57.o
> >> /tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h:
> >> No such file or director
> >>
> >> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> >> -I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> >> /tmp/usession-release-2.3.1-3/platcheck_57.o
> >>
> >>
> >> Thanks,
> >> Joseph David Bor?
> >> josephb.org
> >
> >> _______________________________________________
> >> pypy-dev mailing list
> >> pypy-dev at python.org
> >> https://mail.python.org/mailman/listinfo/pypy-dev
> >
> > _______________________________________________
> > pypy-dev mailing list
> > pypy-dev at python.org
> > https://mail.python.org/mailman/listinfo/pypy-dev

From pjenvey at underboss.org  Fri Jun 20 22:32:08 2014
From: pjenvey at underboss.org (Philip Jenvey)
Date: Fri, 20 Jun 2014 13:32:08 -0700
Subject: [pypy-dev] PyPy3 2.3.1 released
Message-ID: <D566B90F-DAFD-4CE2-821D-A87FB0E998E7@underboss.org>

=====================
PyPy3 2.3.1 - Fulcrum
=====================

We're pleased to announce the first stable release of PyPy3. PyPy3
targets Python 3 (3.2.5) compatibility.

We would like to thank all of the people who donated_ to the `py3k proposal`_
for supporting the work that went into this.

You can download the PyPy3 2.3.1 release here:

    http://pypy.org/download.html#pypy3-2-3-1

Highlights
==========

* The first stable release of PyPy3: support for Python 3!

* The stdlib has been updated to Python 3.2.5

* Additional support for the u'unicode' syntax (`PEP 414`_) from Python 3.3

* Updates from the default branch, such as incremental GC and various JIT
  improvements

* Resolved some notable JIT performance regressions from PyPy2:

 - Re-enabled the previously disabled collection (list/dict/set) strategies

 - Resolved performance of iteration over range objects

 - Resolved handling of Python 3's exception __context__ unnecessarily forcing
   frame object overhead

.. _`PEP 414`: http://legacy.python.org/dev/peps/pep-0414/

What is PyPy?
==============

PyPy is a very compliant Python interpreter, almost a drop-in replacement for
CPython 2.7.6 or 3.2.5. It's fast due to its integrated tracing JIT compiler.

This release supports x86 machines running Linux 32/64, Mac OS X 64, Windows,
and OpenBSD,
as well as newer ARM hardware (ARMv6 or ARMv7, with VFPv3) running Linux.

While we support 32 bit python on Windows, work on the native Windows 64
bit python is still stalling, we would welcome a volunteer
to `handle that`_.

.. _`handle that`: http://doc.pypy.org/en/latest/windows.html#what-is-missing-for-a-full-64-bit-translation

How to use PyPy?
=================

We suggest using PyPy from a `virtualenv`_. Once you have a virtualenv
installed, you can follow instructions from `pypy documentation`_ on how
to proceed. This document also covers other `installation schemes`_.

.. _donated: http://morepypy.blogspot.com/2012/01/py3k-and-numpy-first-stage-thanks-to.html
.. _`py3k proposal`: http://pypy.org/py3donate.html
.. _`pypy documentation`: http://doc.pypy.org/en/latest/getting-started.html#installing-using-virtualenv
.. _`virtualenv`: http://www.virtualenv.org/en/latest/
.. _`installation schemes`: http://doc.pypy.org/en/latest/getting-started.html#installing-pypy


Cheers,
the PyPy team

--
Philip Jenvey


From arigo at tunes.org  Sat Jun 21 14:26:01 2014
From: arigo at tunes.org (Armin Rigo)
Date: Sat, 21 Jun 2014 14:26:01 +0200
Subject: [pypy-dev] how can a class method become frozen in tests only
	on windows?
In-Reply-To: <53A09E90.5080502@gmail.com>
References: <53A09E90.5080502@gmail.com>
Message-ID: <CAMSv6X0ff=CQEBt=nyT=7v2OBumGC6fxKT6x+0nYLt_4XrAoTQ@mail.gmail.com>

Hi Matti,

On 17 June 2014 22:01, Matti Picus <matti.picus at gmail.com> wrote:
> in looking at the own test_rzipfile failures on windows[0], the annotator is
> failing in rtyper.rpbc.redispatch_call since the method being called is a
> MultipleFrozenPBCRepr .

Not sure, but this seems to fix the problem: http://bpaste.net/show/393277/

If you can confirm, feel free to check it in.


A bient?t,

Armin.

From matti.picus at gmail.com  Sat Jun 21 22:24:24 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Sat, 21 Jun 2014 23:24:24 +0300
Subject: [pypy-dev] pythonapi and pypy on windows
Message-ID: <53A5E9F8.1000508@gmail.com>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140621/c798b359/attachment.html>

From alex.gaynor at gmail.com  Sat Jun 21 22:28:44 2014
From: alex.gaynor at gmail.com (Alex Gaynor)
Date: Sat, 21 Jun 2014 13:28:44 -0700
Subject: [pypy-dev] pythonapi and pypy on windows
In-Reply-To: <53A5E9F8.1000508@gmail.com>
References: <53A5E9F8.1000508@gmail.com>
Message-ID: <CAFRnB2WL+W1+LZ1k0=YqhyNGrg1mR6Nof4=GrSrB8n00Q2ZT8w@mail.gmail.com>

I think we shoudl either disable it, or make it be really good: meaning
completely skipping the C-call layer, and just making things super simple:

def PyListAppend(list, item):
    list.append(item)

There's no point in having it if it's super expensive (IMO).

Alex


On Sat, Jun 21, 2014 at 1:24 PM, Matti Picus <matti.picus at gmail.com> wrote:

>  Cpython has a feature that allows you to call c-api functions from python
> via ctypes.pythonapi
> It doesn't currently work on pypy, I started to document why not on the
> disable_pythonapi branch:
>
>     ``ctypes.pythonapi`` lets you access the CPython C API
>     emulation layer. It does not work on PyPy at the moment, we are
> missing a
>     ``getfunc`` method for CDLL. Work was begun
>     to refactor the rpython implementation of _rawffi (in
>     pypy/modules/_rawffi/alt) but that project has stalled.
>
>     Note that even if it worked, our implementation would not do anything
>     sensible about the GIL and the functions will be named with an extra
>     "Py", for example ``PyPyInt_FromLong()``.  Basically, don't use this.
>     Assuming the PyObject pointers you get have any particular fields in
>     any particular order is just going to crash.
>
> but since it is such a bad idea, can we just disable it instead of fixing?
> In Favor of disabling: no need to extend _rawffi, only remove some code
> and fix tests
> Against disabling: another obscure cpython incompatability
>
> Matti
>
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
>
>


-- 
"I disapprove of what you say, but I will defend to the death your right to
say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
"The people's good is the highest law." -- Cicero
GPG Key fingerprint: 125F 5C67 DFE9 4084
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140621/5f7277d3/attachment.html>

From laurie at tratt.net  Sun Jun 22 12:12:08 2014
From: laurie at tratt.net (Laurence Tratt)
Date: Sun, 22 Jun 2014 11:12:08 +0100
Subject: [pypy-dev] 2.3.1 distribution issues?
Message-ID: <20140622101208.GA7266@overdrive.tratt.net>

I feel I might be going mad but I think there are a few web-page issues with
the 2.3.1 release. First the links to pypy-2.3.1-src.tar.bz2 and
pypy-2.3.1-src.zip point to release-2.3.1.tar.bz2 and release-2.3.1.zip
respectively in the get directory [1]. This confused the life out of me for a
while. It also makes OS packagers nervous because there is a history of files
being fetched from tags changing their checksums over time.

[As an aside, decompressing release-2.3.1.tar.bz2 leads to a
pypy-pypy-<somerandomhash> directory, which makes it annoying for OS
packagers to upgrade the package. 2.3 has this problem as well, although I'm
sure some past source releases had a more standard pypy-V.V directory.]

If I go instead to BitBucket's downloads page [2], I can see binaries for
various platforms for 2.3.1, but not a source archive?


Laurie

[1] http://pypy.org/download.html#building-from-source
[2] https://bitbucket.org/pypy/pypy/downloads

From matti.picus at gmail.com  Sun Jun 22 12:27:02 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Sun, 22 Jun 2014 13:27:02 +0300
Subject: [pypy-dev] 2.3.1 distribution issues?
In-Reply-To: <20140622101208.GA7266@overdrive.tratt.net>
References: <20140622101208.GA7266@overdrive.tratt.net>
Message-ID: <53A6AF76.1090604@gmail.com>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140622/5c150b88/attachment.html>

From laurie at tratt.net  Sun Jun 22 18:26:04 2014
From: laurie at tratt.net (Laurence Tratt)
Date: Sun, 22 Jun 2014 17:26:04 +0100
Subject: [pypy-dev] 2.3.1 distribution issues?
In-Reply-To: <53A6AF76.1090604@gmail.com>
References: <20140622101208.GA7266@overdrive.tratt.net>
 <53A6AF76.1090604@gmail.com>
Message-ID: <20140622162604.GD7266@overdrive.tratt.net>

On Sun, Jun 22, 2014 at 01:27:02PM +0300, Matti Picus wrote:

> Sorry, it's my fault, I didn't download/repackage/upload the source files
> as in previous builds, it seemed redundant.
> I see it was not, so I will try to fix.

Thanks Matti!


Laurie

From arigo at tunes.org  Mon Jun 23 09:34:06 2014
From: arigo at tunes.org (Armin Rigo)
Date: Mon, 23 Jun 2014 09:34:06 +0200
Subject: [pypy-dev] pythonapi and pypy on windows
In-Reply-To: <53A5E9F8.1000508@gmail.com>
References: <53A5E9F8.1000508@gmail.com>
Message-ID: <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>

Hi Matti,

First question, what is the relationship of this mail and "PyPy on Windows"?

On 21 June 2014 22:24, Matti Picus <matti.picus at gmail.com> wrote:
>     ``ctypes.pythonapi`` lets you access the CPython C API
>     emulation layer. It does not work on PyPy at the moment, we are missing
> a
>     ``getfunc`` method for CDLL. Work was begun
>     to refactor the rpython implementation of _rawffi (in
>     pypy/modules/_rawffi/alt) but that project has stalled.

I thought we simply didn't provide ``ctypes.pythonapi``.  Seems we do,
but I don't want to know what nonsense it does.  Just remove it?  As
Alex says, I don't think it's as simple as adding some method on some
CDLL.


A bient?t,

Armin.

From piotr.skamruk at gmail.com  Mon Jun 23 10:21:43 2014
From: piotr.skamruk at gmail.com (Piotr Skamruk)
Date: Mon, 23 Jun 2014 10:21:43 +0200
Subject: [pypy-dev] [pypy] Adding gcc flags to Makefile
In-Reply-To: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
References: <CAFFYk1UjNAyBgYhmj922=6aQn2SZZjTMD9_cEKO-4bUjh9gfvg@mail.gmail.com>
Message-ID: <CAO_YSZOgoY_tZdR1DzN8YZV+zKyKL6y50zmkRRQq8nhuz8Q+=w@mail.gmail.com>

bzlib is library for bzip2 archives. it's not zlib which You mentioned
in first sentence.

2014-06-19 10:28 GMT+02:00 Joe Bor? <mail at jdborg.com>:
> Hi all,
>
> Because I've had to build zlib away from its normal home, I need to pass -I
> to gcc (LD_LIBRARY_PATH doesn't work).
>
> Setting this -I fixes my build issue:
>
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
> /tmp/usession-release-2.3.1-3/platcheck_57.c:33:19: fatal error: bzlib.h: No
> such file or director
>
> gcc -c -O3 -pthread -fomit-frame-pointer -Wall -Wno-unused
> -I/scratch/root/include /tmp/usession-release-2.3.1-3/platcheck_57.c -o
> /tmp/usession-release-2.3.1-3/platcheck_57.o
>
>
> Thanks,
> Joseph David Bor?
> josephb.org
>
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> https://mail.python.org/mailman/listinfo/pypy-dev
>

From fabiofz at gmail.com  Tue Jun 24 02:11:42 2014
From: fabiofz at gmail.com (Fabio Zadrozny)
Date: Mon, 23 Jun 2014 21:11:42 -0300
Subject: [pypy-dev] PyPy on JVM
Message-ID: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>

Hi PyPy-dev,

I remember there was some work on getting PyPy on JVM... is that still
possible? Can someone point me on how to do it?

Thank you,

Fabio
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140623/ec818ef9/attachment.html>

From kitizz.devside at gmail.com  Tue Jun 24 02:42:18 2014
From: kitizz.devside at gmail.com (Kit Ham)
Date: Tue, 24 Jun 2014 10:42:18 +1000
Subject: [pypy-dev] Question about PyPy Sandboxing
Message-ID: <CAAPN+9tKY0eLwa14G8ZYg-OpH6kYCv0wUm4qunOcPcD1+DUiuQ@mail.gmail.com>

Hi devs,

Can multiple PyPy sandboxes be executed and interacted with from C/C++?

For context, I'm adding plugin support to my application. I'd like people
to be able to write a python module, and call it's functions from the C/C++
in a sandboxed environment.

This is for 2 reasons:
1) So that a plugin can't mess with the user's system
2) So that a plugin's access to other plugins can be controlled

Cheers,

Kit

P.S. I'm currently using PyOtherSide in my Qt app, but I'm realising it may
not be a good solution for plugins.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140624/aabdb107/attachment.html>

From fijall at gmail.com  Tue Jun 24 09:06:32 2014
From: fijall at gmail.com (Maciej Fijalkowski)
Date: Tue, 24 Jun 2014 09:06:32 +0200
Subject: [pypy-dev] PyPy on JVM
In-Reply-To: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>
References: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>
Message-ID: <CAK5idxQzrDqHxu1HiJt6-gSVsFCFe63LoQyh3hogjpzWcDbhwg@mail.gmail.com>

On Tue, Jun 24, 2014 at 2:11 AM, Fabio Zadrozny <fabiofz at gmail.com> wrote:
> Hi PyPy-dev,
>
> I remember there was some work on getting PyPy on JVM... is that still
> possible? Can someone point me on how to do it?
>
> Thank you,
>
> Fabio

Hi Fabio.

The PyPy JVM backend was abandoned together with ootype.

From arigo at tunes.org  Tue Jun 24 09:09:52 2014
From: arigo at tunes.org (Armin Rigo)
Date: Tue, 24 Jun 2014 09:09:52 +0200
Subject: [pypy-dev] PyPy on JVM
In-Reply-To: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>
References: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>
Message-ID: <CAMSv6X1y-xb3HDqye51t2QD60X5DtQo0V7=ghA0nbpXfN3SWZQ@mail.gmail.com>

Hi Fabio,

On 24 June 2014 02:11, Fabio Zadrozny <fabiofz at gmail.com> wrote:
> I remember there was some work on getting PyPy on JVM... is that still
> possible? Can someone point me on how to do it?

Sorry, we dropped support for that due to general lack of interest.
It was theoretically possible since long ago, but unfinished.  No-one
ever did more than realize (again and again) that it was going to be
more work than they originally thought, notably to integrate the
Python side and the Java side a la Jython.  Then there would have been
quite a lot of additional work to make the JIT work there.  So we
dropped "ootype" support.  This did (and still does) let us simplify
large parts of RPython.


A bient?t,

Armin.

From arigo at tunes.org  Tue Jun 24 09:17:28 2014
From: arigo at tunes.org (Armin Rigo)
Date: Tue, 24 Jun 2014 09:17:28 +0200
Subject: [pypy-dev] Question about PyPy Sandboxing
In-Reply-To: <CAAPN+9tKY0eLwa14G8ZYg-OpH6kYCv0wUm4qunOcPcD1+DUiuQ@mail.gmail.com>
References: <CAAPN+9tKY0eLwa14G8ZYg-OpH6kYCv0wUm4qunOcPcD1+DUiuQ@mail.gmail.com>
Message-ID: <CAMSv6X2p390vQLu=Vs-y_-p-7gxUfsz-CAER0GcvR9YERf2Vgw@mail.gmail.com>

Hi Kit,

On 24 June 2014 02:42, Kit Ham <kitizz.devside at gmail.com> wrote:
> Can multiple PyPy sandboxes be executed and interacted with from C/C++?

Yes, you can in theory do whatever you want, but it is not really
finished and documented.

You first need to decide if it's ok that each sandbox runs in a
separate process or if you would rather have an in-process solution
(more work).  In the first case, you'd extend the "controller"
(regular Python code controlling the sandboxed subprocess) so that it
can handle several subprocesses at once.  Then, for example, you'd run
the controller in CPython, and link to CPython from your C/C++ code
using the standard CPython ways.  Or you can even rewrite the
controller in C/C++: it's just some code reading and writing to the
subprocesses' stdin/stdout.  (Each subprocess is one sandboxed PyPy
instance.)


A bient?t,

Armin.

From fabiofz at gmail.com  Tue Jun 24 12:50:35 2014
From: fabiofz at gmail.com (Fabio Zadrozny)
Date: Tue, 24 Jun 2014 07:50:35 -0300
Subject: [pypy-dev] PyPy on JVM
In-Reply-To: <CAMSv6X1y-xb3HDqye51t2QD60X5DtQo0V7=ghA0nbpXfN3SWZQ@mail.gmail.com>
References: <CANXBEFpY0itv=yqoD3w+aTdoqtp5EG2hyrP8voa3xMtozmK+rg@mail.gmail.com>
 <CAMSv6X1y-xb3HDqye51t2QD60X5DtQo0V7=ghA0nbpXfN3SWZQ@mail.gmail.com>
Message-ID: <CANXBEFq=YiW8_ycK45dV1pJW2oxzUa2YQhXGkYmmvASUoYG9jw@mail.gmail.com>

Ok,

Thank you for the info.

Best Regards,

Fabio


On Tue, Jun 24, 2014 at 4:09 AM, Armin Rigo <arigo at tunes.org> wrote:

> Hi Fabio,
>
> On 24 June 2014 02:11, Fabio Zadrozny <fabiofz at gmail.com> wrote:
> > I remember there was some work on getting PyPy on JVM... is that still
> > possible? Can someone point me on how to do it?
>
> Sorry, we dropped support for that due to general lack of interest.
> It was theoretically possible since long ago, but unfinished.  No-one
> ever did more than realize (again and again) that it was going to be
> more work than they originally thought, notably to integrate the
> Python side and the Java side a la Jython.  Then there would have been
> quite a lot of additional work to make the JIT work there.  So we
> dropped "ootype" support.  This did (and still does) let us simplify
> large parts of RPython.
>
>
> A bient?t,
>
> Armin.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140624/c87ea039/attachment.html>

From matti.picus at gmail.com  Wed Jun 25 14:47:06 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Wed, 25 Jun 2014 15:47:06 +0300
Subject: [pypy-dev] pythonapi
In-Reply-To: <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
References: <53A5E9F8.1000508@gmail.com>
 <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
Message-ID: <53AAC4CA.6060602@gmail.com>

(previous subject mistakenly included reference to "PyPy on Windows")
I created a branch disable_pythonapi that copletely removes pythonapi 
and PyDLL from ctypes,
removes sys.dllhandle on windows, and documents these differences to 
cpython.
Should I merge it to default?
Matti

On 23/06/2014 10:34 AM, Armin Rigo wrote:
> Hi Matti,
>
> First question, what is the relationship of this mail and "PyPy on Windows"?
>
> On 21 June 2014 22:24, Matti Picus <matti.picus at gmail.com> wrote:
>>      ``ctypes.pythonapi`` lets you access the CPython C API
>>      emulation layer. It does not work on PyPy at the moment, we are missing
>> a
>>      ``getfunc`` method for CDLL. Work was begun
>>      to refactor the rpython implementation of _rawffi (in
>>      pypy/modules/_rawffi/alt) but that project has stalled.
> I thought we simply didn't provide ``ctypes.pythonapi``.  Seems we do,
> but I don't want to know what nonsense it does.  Just remove it?  As
> Alex says, I don't think it's as simple as adding some method on some
> CDLL.
>
>
> A bient?t,
>
> Armin.


From arigo at tunes.org  Wed Jun 25 16:04:05 2014
From: arigo at tunes.org (Armin Rigo)
Date: Wed, 25 Jun 2014 16:04:05 +0200
Subject: [pypy-dev] pythonapi
In-Reply-To: <53AAC4CA.6060602@gmail.com>
References: <53A5E9F8.1000508@gmail.com>
 <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
 <53AAC4CA.6060602@gmail.com>
Message-ID: <CAMSv6X175adHjiP1DDhwRp0NCLQAxH9vtWz8vi=Nf3KDiGyduQ@mail.gmail.com>

Hi Matti,

On 25 June 2014 14:47, Matti Picus <matti.picus at gmail.com> wrote:
> removes sys.dllhandle on windows,

Can you explain why removing sys.dllhandle is a good idea?  It seems
unrelated to the ctypes changes, but maybe it is not.


A bient?t,

Armin.

From matti.picus at gmail.com  Wed Jun 25 16:10:24 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Wed, 25 Jun 2014 17:10:24 +0300
Subject: [pypy-dev] pythonapi
In-Reply-To: <CAMSv6X175adHjiP1DDhwRp0NCLQAxH9vtWz8vi=Nf3KDiGyduQ@mail.gmail.com>
References: <53A5E9F8.1000508@gmail.com>
 <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
 <53AAC4CA.6060602@gmail.com>
 <CAMSv6X175adHjiP1DDhwRp0NCLQAxH9vtWz8vi=Nf3KDiGyduQ@mail.gmail.com>
Message-ID: <53AAD850.20001@gmail.com>

Assuming I understand correctly, the reason sys.dllhandle exists is that 
you can do
dlopen(NULL)
in linux to get a handle to the current process, but not
LoadLibrary(NULL).
So windows builds provide a handle to the current process DLL as 
sys.dllhandle,
which is then wrapped into ctypes.PyDLL(..., sys.dllhandle) and served 
up as ctypes.pythonapi.
Working backwards from no pythonapi -> no PyDLL -> no sys.dllhandle.
Matti

On 25/06/2014 5:04 PM, Armin Rigo wrote:
> Hi Matti,
>
> On 25 June 2014 14:47, Matti Picus <matti.picus at gmail.com> wrote:
>> removes sys.dllhandle on windows,
> Can you explain why removing sys.dllhandle is a good idea?  It seems
> unrelated to the ctypes changes, but maybe it is not.
>
>
> A bient?t,
>
> Armin.


From arigo at tunes.org  Wed Jun 25 16:33:06 2014
From: arigo at tunes.org (Armin Rigo)
Date: Wed, 25 Jun 2014 16:33:06 +0200
Subject: [pypy-dev] pythonapi
In-Reply-To: <53AAD850.20001@gmail.com>
References: <53A5E9F8.1000508@gmail.com>
 <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
 <53AAC4CA.6060602@gmail.com>
 <CAMSv6X175adHjiP1DDhwRp0NCLQAxH9vtWz8vi=Nf3KDiGyduQ@mail.gmail.com>
 <53AAD850.20001@gmail.com>
Message-ID: <CAMSv6X3cAfUzK+WJ20wbSE3nGap8t92j08wJt8y5oKT4tnH1Gw@mail.gmail.com>

Hi Matti,

On 25 June 2014 16:10, Matti Picus <matti.picus at gmail.com> wrote:
> Working backwards from no pythonapi -> no PyDLL -> no sys.dllhandle.

Right, but I'm not sure that it's only used by ctypes and nothing
else.  If that were the case it would be some private name in the
_ctypes module...

Also, I'm not sure I understand it correctly, but it seems that on
CPython, sys.dllhandle is just an integer, whereas on PyPy so far it
is some W_CDLL instance.  I wouldn't mind fixing it to be like CPython
(then it has nothing to do with cpyext or with _rawffi).


A bient?t,

Armin.

From yyc1992 at gmail.com  Fri Jun 27 04:09:35 2014
From: yyc1992 at gmail.com (Yichao Yu)
Date: Fri, 27 Jun 2014 10:09:35 +0800
Subject: [pypy-dev] The correct way to incrementally build list in RPython
Message-ID: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>

Hi,

On an attempt to fix the hash function of numpy.dtype, I came up with
the patch in the attachment translated from hashdescr.c in the
original numpy source. The patch passed the test when running in
pyinteractive.py. However, when started to translate pypy, I get an
ListChangeUnallowed error.

I read in the Coding Guide that list.append should be allowed (which
also said there's a lot of special rules) and I have seen usage of
append in other RPython modules. So what is the rule of using
list.append and what is the best way of doing this? (I guess I can
always using application level lists but is that the only/best way?)

Other comments on this patch (style/naming etc) are also welcome.

Yichao Yu
-------------- next part --------------
A non-text attachment was scrubbed...
Name: numpy-hash.patch
Type: text/x-patch
Size: 1712 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20140627/c5658f1e/attachment.bin>

From arigo at tunes.org  Fri Jun 27 09:07:43 2014
From: arigo at tunes.org (Armin Rigo)
Date: Fri, 27 Jun 2014 09:07:43 +0200
Subject: [pypy-dev] The correct way to incrementally build list in
	RPython
In-Reply-To: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
References: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
Message-ID: <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>

Hi Yichao,

On 27 June 2014 04:09, Yichao Yu <yyc1992 at gmail.com> wrote:
> On an attempt to fix the hash function of numpy.dtype, I came up with
> the patch in the attachment translated from hashdescr.c in the
> original numpy source. The patch passed the test when running in
> pyinteractive.py. However, when started to translate pypy, I get an
> ListChangeUnallowed error.

That errors comes from space.newtuple(lst): it must be passed a list
"lst" that is "not resized", that is, a list on which we never do an
append() or pop() or similar.

The standard fix is to do "lst = [None] * length" and then fill
"lst[index] = ...".  This only works if you know the final length in
advance.  A more heavy way to fix it is simply to call
space.newtuple(lst[:]), making a fresh copy of the list.  But in this
case, a better fix would be to avoid the list and the space.newtuple()
and space.hash() completely, and instead compute the hash value
incrementally, by copying the logic in
pypy/objspace/std/tupleobject.py:descr_hash().

Similarly, I'm not sure you need to wrap self.kind, 0, self.elsize and
self.alignment and put them inside a tuple of length 4, when all you
could do is some random computation with these three numbers.


A bient?t,

Armin.

From yyc1992 at gmail.com  Fri Jun 27 09:27:28 2014
From: yyc1992 at gmail.com (Yichao Yu)
Date: Fri, 27 Jun 2014 15:27:28 +0800
Subject: [pypy-dev] The correct way to incrementally build list in
	RPython
In-Reply-To: <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>
References: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
 <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>
Message-ID: <CAMvDr+QOG_PyqS9sNo=czJ2EX4Oi09Ee9qUsJQ-i6ss185+_vA@mail.gmail.com>

On Fri, Jun 27, 2014 at 3:07 PM, Armin Rigo <arigo at tunes.org> wrote:
> Hi Yichao,
>
> On 27 June 2014 04:09, Yichao Yu <yyc1992 at gmail.com> wrote:
>> On an attempt to fix the hash function of numpy.dtype, I came up with
>> the patch in the attachment translated from hashdescr.c in the
>> original numpy source. The patch passed the test when running in
>> pyinteractive.py. However, when started to translate pypy, I get an
>> ListChangeUnallowed error.
>
> That errors comes from space.newtuple(lst): it must be passed a list
> "lst" that is "not resized", that is, a list on which we never do an
> append() or pop() or similar.

Hmmm. So the error actually happens when the compiler traces back
function calls? I was trying to figure out what's wrong from inside my
function T_T ...

>
> The standard fix is to do "lst = [None] * length" and then fill
> "lst[index] = ...".  This only works if you know the final length in
> advance.  A more heavy way to fix it is simply to call
> space.newtuple(lst[:]), making a fresh copy of the list.  But in this
> case, a better fix would be to avoid the list and the space.newtuple()
> and space.hash() completely, and instead compute the hash value
> incrementally, by copying the logic in
> pypy/objspace/std/tupleobject.py:descr_hash().

Thank you for your advices. I would definitely try that.

>
> Similarly, I'm not sure you need to wrap self.kind, 0, self.elsize and
> self.alignment and put them inside a tuple of length 4, when all you
> could do is some random computation with these three numbers.

I agree. As I said, this is basically copied from the c implementation
and this is what they do over there..... (btw, it is actually a tuple
of length 5 with endian being the second element ;P ).

>
>
> A bient?t,
>
> Armin.

Yichao Yu

From yyc1992 at gmail.com  Fri Jun 27 12:17:22 2014
From: yyc1992 at gmail.com (Yichao Yu)
Date: Fri, 27 Jun 2014 18:17:22 +0800
Subject: [pypy-dev] The correct way to incrementally build list in
	RPython
In-Reply-To: <CAMvDr+QOG_PyqS9sNo=czJ2EX4Oi09Ee9qUsJQ-i6ss185+_vA@mail.gmail.com>
References: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
 <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>
 <CAMvDr+QOG_PyqS9sNo=czJ2EX4Oi09Ee9qUsJQ-i6ss185+_vA@mail.gmail.com>
Message-ID: <CAMvDr+Sv_SZpt3agtBM+Jw0kFoqwDvumOb31DnfWUvVZjrsAYA@mail.gmail.com>

On Fri, Jun 27, 2014 at 3:27 PM, Yichao Yu <yyc1992 at gmail.com> wrote:
> On Fri, Jun 27, 2014 at 3:07 PM, Armin Rigo <arigo at tunes.org> wrote:
>> Hi Yichao,
>>
>> On 27 June 2014 04:09, Yichao Yu <yyc1992 at gmail.com> wrote:
>>> On an attempt to fix the hash function of numpy.dtype, I came up with
>>> the patch in the attachment translated from hashdescr.c in the
>>> original numpy source. The patch passed the test when running in
>>> pyinteractive.py. However, when started to translate pypy, I get an
>>> ListChangeUnallowed error.
>>
>> That errors comes from space.newtuple(lst): it must be passed a list
>> "lst" that is "not resized", that is, a list on which we never do an
>> append() or pop() or similar.
>
> Hmmm. So the error actually happens when the compiler traces back
> function calls? I was trying to figure out what's wrong from inside my
> function T_T ...
>
>>
>> The standard fix is to do "lst = [None] * length" and then fill
>> "lst[index] = ...".  This only works if you know the final length in
>> advance.  A more heavy way to fix it is simply to call
>> space.newtuple(lst[:]), making a fresh copy of the list.  But in this
>> case, a better fix would be to avoid the list and the space.newtuple()
>> and space.hash() completely, and instead compute the hash value
>> incrementally, by copying the logic in
>> pypy/objspace/std/tupleobject.py:descr_hash().
>
> Thank you for your advices. I would definitely try that.

I've made a new version[1] and it seems to work.
I am using the hash function from `compute_hash` instead of the
improved one in tupleobject.py:descr_hash since the length is not
known in advance.

[1] https://bitbucket.org/pypy/pypy/pull-request/242/fix-hash-function-for-complicated/diff

>
>>
>> Similarly, I'm not sure you need to wrap self.kind, 0, self.elsize and
>> self.alignment and put them inside a tuple of length 4, when all you
>> could do is some random computation with these three numbers.

I am still creating a tuple out of these 5 numbers/strings since I
don't want to duplicate the hash functions for all of them. Can the
compiler deal with this case efficiently?

>
> I agree. As I said, this is basically copied from the c implementation
> and this is what they do over there..... (btw, it is actually a tuple
> of length 5 with endian being the second element ;P ).
>
>>
>>
>> A bient?t,
>>
>> Armin.
>
> Yichao Yu

From arigo at tunes.org  Fri Jun 27 12:37:32 2014
From: arigo at tunes.org (Armin Rigo)
Date: Fri, 27 Jun 2014 12:37:32 +0200
Subject: [pypy-dev] The correct way to incrementally build list in
	RPython
In-Reply-To: <CAMvDr+Sv_SZpt3agtBM+Jw0kFoqwDvumOb31DnfWUvVZjrsAYA@mail.gmail.com>
References: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
 <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>
 <CAMvDr+QOG_PyqS9sNo=czJ2EX4Oi09Ee9qUsJQ-i6ss185+_vA@mail.gmail.com>
 <CAMvDr+Sv_SZpt3agtBM+Jw0kFoqwDvumOb31DnfWUvVZjrsAYA@mail.gmail.com>
Message-ID: <CAMSv6X39FTuq-bHxHz3AZgP=n8PnJGi5WVHMajYP8_qTk=QH2w@mail.gmail.com>

Hi Yichao,

On 27 June 2014 12:17, Yichao Yu <yyc1992 at gmail.com> wrote:
> I am still creating a tuple out of these 5 numbers/strings since I
> don't want to duplicate the hash functions for all of them. Can the
> compiler deal with this case efficiently?

If by "deal with this case efficiently" you mean, will the compiler
remove the tuple allocation and inline the 5-elements hash, then it
might just be possible depending on a number of details, but I doubt
it.  Better to be explicit.  I had in mind something like:

    my_hash = num1 * 1000003
    my_hash = (my_hash ^ num2) * 1000003
    my_hash = (my_hash ^ compute_hash(string3))
    my_hash = intmask(my_hash)   # at the end


A bient?t,

Armin.

From yyc1992 at gmail.com  Fri Jun 27 15:17:26 2014
From: yyc1992 at gmail.com (Yichao Yu)
Date: Fri, 27 Jun 2014 21:17:26 +0800
Subject: [pypy-dev] The correct way to incrementally build list in
	RPython
In-Reply-To: <CAMSv6X39FTuq-bHxHz3AZgP=n8PnJGi5WVHMajYP8_qTk=QH2w@mail.gmail.com>
References: <CAMvDr+RQ4SNB2neNxj=qK-A9B5To0Po7Da+XKtKwk_tRuVfZ-Q@mail.gmail.com>
 <CAMSv6X3d64AgcoWHmE+f=0yZSgWAXqPZXAcJszBAZ8o75FhQ9Q@mail.gmail.com>
 <CAMvDr+QOG_PyqS9sNo=czJ2EX4Oi09Ee9qUsJQ-i6ss185+_vA@mail.gmail.com>
 <CAMvDr+Sv_SZpt3agtBM+Jw0kFoqwDvumOb31DnfWUvVZjrsAYA@mail.gmail.com>
 <CAMSv6X39FTuq-bHxHz3AZgP=n8PnJGi5WVHMajYP8_qTk=QH2w@mail.gmail.com>
Message-ID: <CAMvDr+SknBQDPaja5VBfAzdA9E6E3mLOus-YyBiBJGNM+xdwFQ@mail.gmail.com>

On Fri, Jun 27, 2014 at 6:37 PM, Armin Rigo <arigo at tunes.org> wrote:
> Hi Yichao,
>
> On 27 June 2014 12:17, Yichao Yu <yyc1992 at gmail.com> wrote:
>> I am still creating a tuple out of these 5 numbers/strings since I
>> don't want to duplicate the hash functions for all of them. Can the
>> compiler deal with this case efficiently?
>
> If by "deal with this case efficiently" you mean, will the compiler
> remove the tuple allocation and inline the 5-elements hash, then it
> might just be possible depending on a number of details, but I doubt
> it.  Better to be explicit.  I had in mind something like:
>
>     my_hash = num1 * 1000003
>     my_hash = (my_hash ^ num2) * 1000003
>     my_hash = (my_hash ^ compute_hash(string3))
>     my_hash = intmask(my_hash)   # at the end
>

Done

>
> A bient?t,
>
> Armin.

From arigo at tunes.org  Sat Jun 28 17:10:35 2014
From: arigo at tunes.org (Armin Rigo)
Date: Sat, 28 Jun 2014 17:10:35 +0200
Subject: [pypy-dev] europython sprints?
In-Reply-To: <CAMSv6X0fa6fYVfMFC_e7BqOHrz+dSuTwsbPm0gXimTKmHaNfJA@mail.gmail.com>
References: <CAH88djX31ERdmg=J_hEydHwg+_jxez7g1VHpH2gLueWJJ5vr=A@mail.gmail.com>
 <CAMSv6X0fa6fYVfMFC_e7BqOHrz+dSuTwsbPm0gXimTKmHaNfJA@mail.gmail.com>
Message-ID: <CAMSv6X2ut36onFDnmz9mYo6vRK9g-YtwLAMzFMFid0hcrEDU7g@mail.gmail.com>

Hi all,

On 3 June 2014 09:56, Armin Rigo <arigo at tunes.org> wrote:
> I guess there will be a PyPy sprint, yes.

It seems there will be an informal PyPy sprint on the first sprint day
(Saturday 26), with at least Romain, Manuel Jacob, myself, and
probably a few others.


A bient?t,

Armin.

From matti.picus at gmail.com  Mon Jun 30 03:12:07 2014
From: matti.picus at gmail.com (Matti Picus)
Date: Mon, 30 Jun 2014 11:12:07 +1000
Subject: [pypy-dev] pythonapi
In-Reply-To: <CAMSv6X3cAfUzK+WJ20wbSE3nGap8t92j08wJt8y5oKT4tnH1Gw@mail.gmail.com>
References: <53A5E9F8.1000508@gmail.com>
 <CAMSv6X2s4cNY_1sLYyG1VqHmxFmdsNyoXtrWZY9PHtAL5bu2ug@mail.gmail.com>
 <53AAC4CA.6060602@gmail.com>
 <CAMSv6X175adHjiP1DDhwRp0NCLQAxH9vtWz8vi=Nf3KDiGyduQ@mail.gmail.com>
 <53AAD850.20001@gmail.com>
 <CAMSv6X3cAfUzK+WJ20wbSE3nGap8t92j08wJt8y5oKT4tnH1Gw@mail.gmail.com>
Message-ID: <53B0B967.4050309@gmail.com>

Googling sys.dllhandle gave me no other uses, the only use case I can of 
think would
involve casting to some kind of ctypes dll and accessing capi (cpyext) 
capabilites.
The original failing test that led me down the path to discover all this 
was
module.cpyext.test.test_cpyext's   AppTestApi.().test_dllhandle - note 
it is in
cpyext not sys, there are no tests in sys in pypy nor in cpython that 
use dllhandle.

However, for the mean time, I have restored sys.dllhandle as an int.

If it seems OK, feel free to merge into default
Matti

On 25/06/2014 5:33 PM, Armin Rigo wrote:
> Hi Matti,
>
> On 25 June 2014 16:10, Matti Picus<matti.picus at gmail.com>  wrote:
>> Working backwards from no pythonapi -> no PyDLL -> no sys.dllhandle.
> Right, but I'm not sure that it's only used by ctypes and nothing
> else.  If that were the case it would be some private name in the
> _ctypes module...
>
> Also, I'm not sure I understand it correctly, but it seems that on
> CPython, sys.dllhandle is just an integer, whereas on PyPy so far it
> is some W_CDLL instance.  I wouldn't mind fixing it to be like CPython
> (then it has nothing to do with cpyext or with _rawffi).
>
>
> A bient?t,
>
> Armin.