1.1.x branch and 1.1.0 tag imminent

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

1.1.x branch and 1.1.0 tag imminent

Jarrod Millman
Hello,

I plan to branch 1.1.x and tag 1.1.0 later today.  As of now, please
consider the trunk in an near absolute freeze.  If you feel that there
is some unimaginably important change that must take place before we
branch and tag, please send an email to the mailing list including
your proposed patch and an explanation of why it is absolutely
necessary.  If the list overwhelmingly agrees with you, I will
consider applying the patch before branching.  If there is any reason
why I shouldn't branch right now, please let me know ASAP!

Thanks to everyone who put so much time and effort into this release.
The trunk has seen tremendous improvements in terms of bug-fixing,
increased testing and improved documentation.  I am looking forward to
a whole host of improvements over the summer.  Once I create the new
branch, I will designate the trunk ready for 1.2 development.  Despite
the increased minor number, please remain cautious with your changes.
As always any changes to the trunk shouldn't break it.  If you have
more experimental work that you want to try, please create a branch.

Thanks,

--
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Charles R Harris


On Fri, May 23, 2008 at 3:31 PM, Jarrod Millman <[hidden email]> wrote:
Hello,

I plan to branch 1.1.x and tag 1.1.0 later today.  As of now, please
consider the trunk in an near absolute freeze.  If you feel that there
is some unimaginably important change that must take place before we
branch and tag, please send an email to the mailing list including
your proposed patch and an explanation of why it is absolutely
necessary.  If the list overwhelmingly agrees with you, I will
consider applying the patch before branching.  If there is any reason
why I shouldn't branch right now, please let me know ASAP!

Thanks to everyone who put so much time and effort into this release.
The trunk has seen tremendous improvements in terms of bug-fixing,
increased testing and improved documentation.  I am looking forward to
a whole host of improvements over the summer.  Once I create the new
branch, I will designate the trunk ready for 1.2 development.  Despite
the increased minor number, please remain cautious with your changes.
As always any changes to the trunk shouldn't break it.  If you have
more experimental work that you want to try, please create a branch.

Will you just kick the d*mn thing out the door?

TIA

Chuck



_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Jarrod Millman
On Fri, May 23, 2008 at 10:14 PM, Charles R Harris
<[hidden email]> wrote:
> Will you just kick the d*mn thing out the door?

I would be happy to my only concern is that I would like to avoid
releasing something that is broken.  Can I safely ignore the 5
buildbot failures that you pointed out yesterday?

======================================================================
FAIL: test_divide (test_errstate.TestErrstate)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "<string>", line 38, in test_divide
AssertionError

======================================================================
FAIL: test_invalid (test_errstate.TestErrstate)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "<string>", line 24, in test_invalid
AssertionError

======================================================================
FAIL: test_divideerr (numpy.core.tests.test_numeric.TestSeterr)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/buildbot/numpy/b2/numpy-install/lib/python2.6/site-packages/numpy/core/tests/test_numeric.py",
line 196, in test_divideerr
    self.fail()
AssertionError

======================================================================
FAIL: test_divide (numpy.core.tests.test_errstate.TestErrstate)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "<string>", line 38, in test_divide
AssertionError

======================================================================
FAIL: test_invalid (numpy.core.tests.test_errstate.TestErrstate)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "<string>", line 24, in test_invalid

AssertionError

If those failures aren't real, I will go ahead and branch.  Sorry that
this process has been so difficult and long.  We can talk about
whether there is a better way to do releases after I finalize 1.1.0.
It may make sense to start rotating release management as David
suggested.

Thanks,

--
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

cdavid
Jarrod Millman wrote:
>
> I would be happy to my only concern is that I would like to avoid
> releasing something that is broken.  Can I safely ignore the 5
> buildbot failures that you pointed out yesterday?
>  

Where do those errors appear ? I don't see them on the builbot. Are they
2.6 specific ? If yes, I would say ignore them, because 2.6 is not
released yet, and is scheduled for september.

cheers,

David
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Jarrod Millman
On Fri, May 23, 2008 at 11:16 PM, David Cournapeau
<[hidden email]> wrote:
> Where do those errors appear ? I don't see them on the builbot. Are they
> 2.6 specific ? If yes, I would say ignore them, because 2.6 is not
> released yet, and is scheduled for september.

You can see them here:
http://buildbot.scipy.org/builders/Linux_x86_64_Fedora/builds/486/steps/shell_2/logs/stdio
http://buildbot.scipy.org/builders/Linux_x86_Fedora_Py2.6/builds/461/steps/shell_2/logs/stdio

You have to search for them, because as Chuck pointed out it seems
valgrind seems to be hiding them.  We should figure out how to avoid
this at some point down the road as well.

I am not sure whether or not this is 2.6 specific or not.

--
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

cdavid
Jarrod Millman wrote:

>
> You can see them here:
> http://buildbot.scipy.org/builders/Linux_x86_64_Fedora/builds/486/steps/shell_2/logs/stdio
> http://buildbot.scipy.org/builders/Linux_x86_Fedora_Py2.6/builds/461/steps/shell_2/logs/stdio
>
> You have to search for them, because as Chuck pointed out it seems
> valgrind seems to be hiding them.  We should figure out how to avoid
> this at some point down the road as well.
>
> I am not sure whether or not this is 2.6 specific or not.
>  

I don't see any error with r5226 on Ubuntu (32 bits). So either it is
fedora specific, 64 bits specific, or python 2.6 specific (or not being
exclusive). Which 2.6 version is used ?

I am testing now with python 2.6 alpha 3 on Ubuntu 32 bits, and I can
also do it in 64 bits. FC would take a lot of time for me, so someone
with Fedora would be more suitable to do it.

I too think we should make the release ASAP, I am tired of it (as you
certainly are :) ).

cheers,

David

_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Jarrod Millman
On Fri, May 23, 2008 at 11:28 PM, David Cournapeau
<[hidden email]> wrote:
> I too think we should make the release ASAP, I am tired of it (as you
> certainly are :) ).

Absolutely!  I just created the 1.1.x maintenance branch:
http://projects.scipy.org/scipy/numpy/changeset/5227

The trunk is now open for 1.2 development:
http://projects.scipy.org/scipy/numpy/changeset/5228
I know several people have code for 1.2 who have been patiently
waiting for me.  Please feel free to go ahead and commit your code to
the trunk now.  Alan McIntyre officially starts his Google Summer of
Code project on Monday.  One of his first tasks will be to help
convert NumPy to the nose testing framework.  That will probably start
in a branch and then get merged all at once just like Matthew Brett
did for SciPy.  Also Stefan should feel free to start committing all
the documentation work to the trunk now.

If no one has any more input on the recent buildbot failures, I will
tag 1.1.0 from the branch tomorrow morning.  Chris Burns is on
vacation (it is memorial day in the States) until Monday, so he won't
be able to create the Mac binaries until then.  Most likely, since
David is around the Windows binary will be available shortly after I
tag the release.

I will announce the NumPy 1.1.0 release late Monday night or early on
Tuesday.  Thanks to everyone who worked so hard getting this release
out.  Sorry that the release management was so uneven and frustrating.
 I will be happy to discuss what could have been done differently or
better.

I have tried to keep the release notes up-to-date as we went; but if
you have any time, now would be a great time for a final pass:
http://projects.scipy.org/scipy/numpy/milestone/1.1.0

Finally, I have also updated Trac so that new tickets default to the
1.2 milestone.  If you create a ticket for 1.1.1, please be sure to
change the milestone from the default.

Thanks,

--
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Charles R Harris


On Sat, May 24, 2008 at 2:48 AM, Jarrod Millman <[hidden email]> wrote:
On Fri, May 23, 2008 at 11:28 PM, David Cournapeau
<[hidden email]> wrote:
> I too think we should make the release ASAP, I am tired of it (as you
> certainly are :) ).

Absolutely!  I just created the 1.1.x maintenance branch:
http://projects.scipy.org/scipy/numpy/changeset/5227

The trunk is now open for 1.2 development:
http://projects.scipy.org/scipy/numpy/changeset/5228
I know several people have code for 1.2 who have been patiently
waiting for me.  Please feel free to go ahead and commit your code to
the trunk now.  Alan McIntyre officially starts his Google Summer of
Code project on Monday.  One of his first tasks will be to help
convert NumPy to the nose testing framework.  That will probably start
in a branch and then get merged all at once just like Matthew Brett
did for SciPy.  Also Stefan should feel free to start committing all
the documentation work to the trunk now.

If no one has any more input on the recent buildbot failures, I will
tag 1.1.0 from the branch tomorrow morning.  Chris Burns is on
vacation (it is memorial day in the States) until Monday, so he won't
be able to create the Mac binaries until then.  Most likely, since
David is around the Windows binary will be available shortly after I
tag the release.

I will announce the NumPy 1.1.0 release late Monday night or early on
Tuesday.  Thanks to everyone who worked so hard getting this release
out.  Sorry that the release management was so uneven and frustrating.
 I will be happy to discuss what could have been done differently or
better.

I have tried to keep the release notes up-to-date as we went; but if
you have any time, now would be a great time for a final pass:
http://projects.scipy.org/scipy/numpy/milestone/1.1.0

Because the default testing level is now all=True some old tests no longer in numpy are also run if they are hanging around in site-packages. So I think we should have a note telling folks to start with a clean install, i.e., they should delete the previous install from site-packages.

Chuck



_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|

Re: 1.1.x branch and 1.1.0 tag imminent

Anne Archibald
In reply to this post by Jarrod Millman
2008/5/24 Jarrod Millman <[hidden email]>:

> On Fri, May 23, 2008 at 11:16 PM, David Cournapeau
> <[hidden email]> wrote:
>> Where do those errors appear ? I don't see them on the builbot. Are they
>> 2.6 specific ? If yes, I would say ignore them, because 2.6 is not
>> released yet, and is scheduled for september.
>
> You can see them here:
> http://buildbot.scipy.org/builders/Linux_x86_64_Fedora/builds/486/steps/shell_2/logs/stdio
> http://buildbot.scipy.org/builders/Linux_x86_Fedora_Py2.6/builds/461/steps/shell_2/logs/stdio
>
> You have to search for them, because as Chuck pointed out it seems
> valgrind seems to be hiding them.  We should figure out how to avoid
> this at some point down the road as well.
>
> I am not sure whether or not this is 2.6 specific or not.

No. These are a known limitation of valgrind when dealing with
floating-point. Other than emailing the valgrind developers to tell
them that yes, somebody cares about this, I think they can be safely
ignored (thank goodness).

Anne
_______________________________________________
Numpy-discussion mailing list
[hidden email]
http://projects.scipy.org/mailman/listinfo/numpy-discussion