Canopy status / updates / courses at Scipy

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

Canopy status / updates / courses at Scipy

Michael Aye
I am a bit worried that the courses next week are based on Canopy. I
just tried to updated 37 packages and the update system failed, first
because biopython had a md5sum mismatch and then the whole update chain
broke down and I had to manually click-close 30 windows telling me:

The package manager has encountered error
PackageManager is busy. Please wait until the current operation is finished.

So, is there really hope that this works until next week?
Best regards,
Michael



_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Michael Aye
On 2013-06-18 18:48:19 +0000, K.-Michael Aye said:

> I am a bit worried that the courses next week are based on Canopy. I
> just tried to updated 37 packages and the update system failed, first
> because biopython had a md5sum mismatch and then the whole update chain
> broke down and I had to manually click-close 30 windows telling me:
>
> The package manager has encountered error
> PackageManager is busy. Please wait until the current operation is finished.
>
> So, is there really hope that this works until next week?
> Best regards,
> Michael

For completeness: Using 1.0.1.1190 on my 10.8.4 Macbook Pro.


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Jonathan March
Michael, 

Many thanks for the feedback and concern. These are real issues that deserve, and are receiving, serious attention. 

For those readers missing context, Michael is referring to the tutorials at next week's SciPy Conference.

1)  > I am a bit worried that the courses next week are based on Canopy.

Canopy is not a requirement for any of the tutorials AFAIK.
Rather, Enthought is offering Canopy as an option to teachers and students who wish to use it. 
They are also welcome to use whatever Python distribution, package source, and tools they prefer.

Canopy has actually been used in quite a few Enthought training courses recently; the integrated editor, file browser, and ipython prompt provide a comfortable integrated working environment and remove infrastructure headache for newcomers.

2) That said, you are absolutely right that the Package Manager's bulk update commands are not working well at present. IMO this is for two reasons: an unclear and unresponsive UI (see below), and some residual package dependency problems resulting from our recent update to numpy 1.7.1. We will be fixing these dependency problems in our online package repo, as much as possible, before the tutorials.

In the meantime, you should find, as others have, after several cycles of "Update All", that the package collection will be stable and functional.

IMO, the biggest problem with the bulk update UI is the lack of clear progress reporting and queuing -- so if you got 30 separate "... please wait..." messages, it is because you clicked 30 "Install" or "Remove" buttons while the package manager was trying to Install All. 

So it will help to warn those on a slow internet connection to be patient, and look at the admittedly obscure spinning icon in the lower left, while installing multiple packages.

However, the best workaround until the UI is fixed may be that users should install the packages they need individually.

3) It seems that you did not try this, but it is worth noting that the "Install All Canopy Packages" button is subject to the same issues as "Update All..", but at a larger scale.

4) We were not able to reproduce the biopython MD5 error. Most likely a one-time download error. Could you try again and if it persists, send us a screen shot?

Thanks,
Jonathan



On Tue, Jun 18, 2013 at 1:56 PM, K.-Michael Aye <[hidden email]> wrote:
On 2013-06-18 18:48:19 +0000, K.-Michael Aye said:

> I am a bit worried that the courses next week are based on Canopy. I
> just tried to updated 37 packages and the update system failed, first
> because biopython had a md5sum mismatch and then the whole update chain
> broke down and I had to manually click-close 30 windows telling me:
>
> The package manager has encountered error
> PackageManager is busy. Please wait until the current operation is finished.
>
> So, is there really hope that this works until next week?
> Best regards,
> Michael

For completeness: Using 1.0.1.1190 on my 10.8.4 Macbook Pro.


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Michael Aye
Hi Jonathan,

thanks for the fast reply.

On 2013-06-18 20:53:59 +0000, Jonathan March said:

> Michael, 
>
> Many thanks for the feedback and concern. These are real issues that
> deserve, and are receiving, serious attention. 
>
> For those readers missing context, Michael is referring to the
> tutorials at next week's SciPy Conference.
>
> 1)  > I am a bit worried that the courses next week are based on Canopy.
>
> Canopy is not a requirement for any of the tutorials AFAIK.
> Rather, Enthought is offering Canopy as an option to teachers and
> students who wish to use it. 
> They are also welcome to use whatever Python distribution, package
> source, and tools they prefer.
>
> Canopy has actually been used in quite a few Enthought training courses
> recently; the integrated editor, file browser, and ipython prompt
> provide a comfortable integrated working environment and remove
> infrastructure headache for newcomers.
>
> 2) That said, you are absolutely right that the Package Manager's bulk
> update commands are not working well at present. IMO this is for two
> reasons: an unclear and unresponsive UI (see below), and some residual
> package dependency problems resulting from our recent update to numpy
> 1.7.1. We will be fixing these dependency problems in our online
> package repo, as much as possible, before the tutorials.
>
> In the meantime, you should find, as others have, after several cycles
> of "Update All", that the package collection will be stable and
> functional.

Another try of 'Install all updates"  went through without problems up
to the last package, matplotlib, which needed another 'motivational'
click on "Update".

>
> IMO, the biggest problem with the bulk update UI is the lack of clear
> progress reporting and queuing -- so if you got 30 separate "... please
> wait..." messages, it is because you clicked 30 "Install" or "Remove"
> buttons while the package manager was trying to Install All. 

I hope you mean that I 'virtually' clicked on 30 Install or Remove
buttons, because I certainly did not do so literally. I always use
"Install all updates".

My guess at what happened is an unsafe thread situation:
* biopython had a packet loss at download creating the md5 mismatch
* the UI popped up a question/warning to me, but locked the update
process thread that wanted to go on while I was answering the
question/warning.
* each update process then creates another UI element warning me about
the busy updater, still being blocked by the md5sum problem to be
answered by the user.

>
> So it will help to warn those on a slow internet connection to be
> patient, and look at the admittedly obscure spinning icon in the lower
> left, while installing multiple packages.
>
> However, the best workaround until the UI is fixed may be that users
> should install the packages they need individually.

What about the versions numbers that were not different but still were
scheduled for an update?
>
> 3) It seems that you did not try this, but it is worth noting that the
> "Install All Canopy Packages" button is subject to the same issues as
> "Update All..", but at a larger scale.

I would like to do that but that list needs a clean up, there are 2
overlapping packages in there, scikit_learn 0.13.1 and scikits.learn
0.8. Not sure if there are more.

>
> 4) We were not able to reproduce the biopython MD5 error. Most likely a
> one-time download error. Could you try again and if it persists, send
> us a screen shot?

Worked fine at the 2nd try.

Another question for setting up a stable work environment:
What is the sanctioned (if any) procedure to install, for example, the
git head of Pandas on top of my official Canopy environment? Last time
I tried it from command-line with the Canopy python in my path and
using just setup.py install, the pandas version did not get updated,
maybe because the PYTHONPATH was not adapted? This is possibly an issue
or clash with your internal virtual-env'ing going on?

Cheers,
Michael


>
> Thanks,
> Jonathan
>
>
>
> On Tue, Jun 18, 2013 at 1:56 PM, K.-Michael Aye
> <[hidden email]> wrote:
> On 2013-06-18 18:48:19 +0000, K.-Michael Aye said:
>
> > I am a bit worried that the courses next week are based on Canopy. I
> > just tried to updated 37 packages and the update system failed, first
> > because biopython had a md5sum mismatch and then the whole update chain
> > broke down and I had to manually click-close 30 windows telling me:
> >
> > The package manager has encountered error
> > PackageManager is busy. Please wait until the current operation is finished.
> >
> > So, is there really hope that this works until next week?
> > Best regards,
> > Michael
>
> For completeness: Using 1.0.1.1190 on my 10.8.4 Macbook Pro.
>
>
> _______________________________________________
> Enthought-Dev mailing list
> [hidden email]
> https://mail.enthought.com/mailman/listinfo/enthought-dev
>
> _______________________________________________
> Enthought-Dev mailing list
> [hidden email]
> https://mail.enthought.com/mailman/listinfo/enthought-dev



_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Jonathan March

On Tue, Jun 18, 2013 at 4:21 PM, K.-Michael Aye <[hidden email]> wrote:
>
> IMO, the biggest problem with the bulk update UI is the lack of clear
> progress reporting and queuing -- so if you got 30 separate "... please
> wait..." messages, it is because you clicked 30 "Install" or "Remove"
> buttons while the package manager was trying to Install All. 

I hope you mean that I 'virtually' clicked on 30 Install or Remove
buttons, because I certainly did not do so literally. I always use
"Install all updates".

My guess at what happened is an unsafe thread situation:
* biopython had a packet loss at download creating the md5 mismatch
* the UI popped up a question/warning to me, but locked the update
process thread that wanted to go on while I was answering the
question/warning.
* each update process then creates another UI element warning me about
the busy updater, still being blocked by the md5sum problem to be
answered by the user.

Thanks, nice theory; we'll look into this.

 
What about the versions numbers that were not different but still were
scheduled for an update?

This is expected and appropriate. The version number is determined by the package author. If we rebuild for some reason then we ++ the build number but we should not modify the version number. Sometimes a rebuild fixes a build bug. Sometimes it modifies a dependency (e.g. when numpy 1.7.1 was built, then lots of packages changed dependency).

The problem is in the UI that does not make this clear unless one opens up "More Info"
  

> 3) It seems that you did not try this, but it is worth noting that the
> "Install All Canopy Packages" button is subject to the same issues as
> "Update All..", but at a larger scale.

I would like to do that but that list needs a clean up, there are 2
overlapping packages in there, scikit_learn 0.13.1 and scikits.learn
0.8. Not sure if there are more.

Yes, we are aware of this one. It's a little tricky because the old package needs to remain in the repo for legacy users (the same repo supports EPD), but we definitely should be excluding it in Canopy.

 
Another question for setting up a stable work environment:
What is the sanctioned (if any) procedure to install, for example, the
git head of Pandas on top of my official Canopy environment? Last time
I tried it from command-line with the Canopy python in my path and
using just setup.py install, the pandas version did not get updated,
maybe because the PYTHONPATH was not adapted? This is possibly an issue
or clash with your internal virtual-env'ing going on?

With PATH correct, such installation should work normally.

It is likely that you had the wrong Canopy Python in your Path (either Canopy base python or System venv python, instead of the User venv.)

This can happen if, on installation, one does not accept the default "Make Canopy my default Python", and then makes a wrong assumption when adding it later manually. In Canopy 1.0.3 the installer will provide a warning about this.

Meanwhile, please see:


** Back to the Package Manager, following up on my earlier comment, "some residual package dependency problems resulting from our recent update to numpy 1.7.1."

We are actually not finding such dependency problems. Could you be so kind as to help by sending us (to [hidden email]) the enpkg / Package Manager log? 

From a terminal prompt where the Canopy User Python is default (as described in the first support article above):
enpkg --config
enpkg --log


Thanks!
Jonathan


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Jonathan March



On Wed, Jun 19, 2013 at 1:06 PM, Jonathan March <[hidden email]> wrote:

On Tue, Jun 18, 2013 at 4:21 PM, K.-Michael Aye <[hidden email]> wrote:

What about the versions numbers that were not different but still were
scheduled for an update?

This is expected and appropriate. The version number is determined by the package author. If we rebuild for some reason then we ++ the build number but we should not modify the version number. Sometimes a rebuild fixes a build bug. Sometimes it modifies a dependency (e.g. when numpy 1.7.1 was built, then lots of packages changed dependency).

The problem is in the UI that does not make this clear unless one opens up "More Info"

Actually, not even there! Sorry, don't know when this vanished, but IMO this info is crucial.


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Michael Aye
In reply to this post by Jonathan March
On 2013-06-19 18:06:41 +0000, Jonathan March said:

>
> On Tue, Jun 18, 2013 at 4:21 PM, K.-Michael Aye <[hidden email]> wrote:
> >
> > IMO, the biggest problem with the bulk update UI is the lack of clear
> > progress reporting and queuing -- so if you got 30 separate "... please
> > wait..." messages, it is because you clicked 30 "Install" or "Remove"
> > buttons while the package manager was trying to Install All. 
>
> I hope you mean that I 'virtually' clicked on 30 Install or Remove
> buttons, because I certainly did not do so literally. I always use
> "Install all updates".
>
> My guess at what happened is an unsafe thread situation:
> * biopython had a packet loss at download creating the md5 mismatch
> * the UI popped up a question/warning to me, but locked the update
> process thread that wanted to go on while I was answering the
> question/warning.
> * each update process then creates another UI element warning me about
> the busy updater, still being blocked by the md5sum problem to be
> answered by the user.
>
> Thanks, nice theory; we'll look into this.
>
>  
> What about the versions numbers that were not different but still were
> scheduled for an update?
>
> This is expected and appropriate. The version number is determined by
> the package author. If we rebuild for some reason then we ++ the build
> number but we should not modify the version number. Sometimes a rebuild
> fixes a build bug. Sometimes it modifies a dependency (e.g. when numpy
> 1.7.1 was built, then lots of packages changed dependency).
>
> The problem is in the UI that does not make this clear unless one opens
> up "More Info"
>   
>
> > 3) It seems that you did not try this, but it is worth noting that the
> > "Install All Canopy Packages" button is subject to the same issues as
> > "Update All..", but at a larger scale.
>
> I would like to do that but that list needs a clean up, there are 2
> overlapping packages in there, scikit_learn 0.13.1 and scikits.learn
> 0.8. Not sure if there are more.
>
> Yes, we are aware of this one. It's a little tricky because the old
> package needs to remain in the repo for legacy users (the same repo
> supports EPD), but we definitely should be excluding it in Canopy.
>
>  
>
> Another question for setting up a stable work environment:
> What is the sanctioned (if any) procedure to install, for example, the
> git head of Pandas on top of my official Canopy environment? Last time
> I tried it from command-line with the Canopy python in my path and
> using just setup.py install, the pandas version did not get updated,
> maybe because the PYTHONPATH was not adapted? This is possibly an issue
> or clash with your internal virtual-env'ing going on?
>
> With PATH correct, such installation should work normally.
>
> It is likely that you had the wrong Canopy Python in your Path (either
> Canopy base python or System venv python, instead of the User venv.)

I think that was the issue for me. I know have a pandas head installed
in Canopy, so that's nice.

>
> This can happen if, on installation, one does not accept the default
> "Make Canopy my default Python", and then makes a wrong assumption when
> adding it later manually. In Canopy 1.0.3 the installer will provide a
> warning about this.
>
> Meanwhile, please see:
> https://support.enthought.com/entries/23646538-Make-Canopy-s-Python-be-your-default-Python-i-e-on-the-PATH- 
>
> https://support.enthought.com/entries/23389761-Installing-packages-into-Canopy-Python-from-the-command-line 
>
>
>
> ** Back to the Package Manager, following up on my earlier comment,
> "some residual package dependency problems resulting from our recent
> update to numpy 1.7.1."
>
> We are actually not finding such dependency problems. Could you be so
> kind as to help by sending us (to
> [hidden email]) the enpkg /
> Package Manager log? 

I don't have any issues with numpy and did not bring it up either, so
I'm not sure how my log will help you? ;)


More seriously, I just found out that I cannot launch any notebook!
I am trying the usual way with
ipython notebook
the notebook desktop (I call it) comes up without probs, but launching
any notebook or creating a new one gives me websocket errors:

A WebSocket connection to could not be established. You will NOT be
able to run code. Check your network connection or notebook server
configuration

Is there a solution for that?

Michael


>
> From a terminal prompt where the Canopy User Python is default (as
> described in the first support article above):
> enpkg --config
> enpkg --log
>
>
> Thanks!
> Jonathan
>
> _______________________________________________
> Enthought-Dev mailing list
> [hidden email]
> https://mail.enthought.com/mailman/listinfo/enthought-dev



_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Jonathan March



On Wed, Jun 19, 2013 at 1:55 PM, K.-Michael Aye <[hidden email]> wrote:
On 2013-06-19 18:06:41 +0000, Jonathan March said:
 
 
> ** Back to the Package Manager, following up on my earlier comment,
> "some residual package dependency problems resulting from our recent
> update to numpy 1.7.1."
>
> We are actually not finding such dependency problems. Could you be so
> kind as to help by sending us (to
> [hidden email]) the enpkg /
> Package Manager log? 

I don't have any issues with numpy and did not bring it up either, so
I'm not sure how my log will help you? ;)

It might not. I probably read more into your original message than was actually there ;) 
Before writing your original message, had you needed to "Install all updates" more than once in a short time?
If so, your log may be of help. 



More seriously, I just found out that I cannot launch any notebook!
I am trying the usual way with
ipython notebook
the notebook desktop (I call it) comes up without probs, but launching
any notebook or creating a new one gives me websocket errors:

A WebSocket connection to could not be established. You will NOT be
able to run code. Check your network connection or notebook server
configuration

Is there a solution for that?

I have not heard of this problem. Are you running ipython from master? Or from somewhere other than the Canopy User Python?
 

Michael


>
> From a terminal prompt where the Canopy User Python is default (as
> described in the first support article above):
> enpkg --config
> enpkg --log
>
>
> Thanks!
> Jonathan
>
> _______________________________________________
> Enthought-Dev mailing list
> [hidden email]
> https://mail.enthought.com/mailman/listinfo/enthought-dev



_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev


_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Michael Aye
On 2013-06-19 20:39:39 +0000, Jonathan March said:

>
>
>
> On Wed, Jun 19, 2013 at 1:55 PM, K.-Michael Aye <[hidden email]> wrote:
> On 2013-06-19 18:06:41 +0000, Jonathan March said:
>  
>  
> > ** Back to the Package Manager, following up on my earlier comment,
> > "some residual package dependency problems resulting from our recent
> > update to numpy 1.7.1."
> >
> > We are actually not finding such dependency problems. Could you be so
> > kind as to help by sending us (to
> > [hidden email]) the enpkg /
> > Package Manager log? 
>
> I don't have any issues with numpy and did not bring it up either, so
> I'm not sure how my log will help you? ;)
>
> It might not. I probably read more into your original message than was
> actually there ;) 
> Before writing your original message, had you needed to "Install all
> updates" more than once in a short time?
> If so, your log may be of help. 

Attached below.

>
>
>
> More seriously, I just found out that I cannot launch any notebook!
> I am trying the usual way with
> ipython notebook
> the notebook desktop (I call it) comes up without probs, but launching
> any notebook or creating a new one gives me websocket errors:
>
> A WebSocket connection to could not be established. You will NOT be
> able to run code. Check your network connection or notebook server
> configuration
>
> Is there a solution for that?
>
> I have not heard of this problem. Are you running ipython from master?
> Or from somewhere other than the Canopy User Python?
>  
I am using '1.0.dev' which I believe was delivered with Canopy?

Brett Murphy created a support case for me (# 10578)

Michael


# config and log infos


[maye@lunatic ~]$ enpkg --config
Python version: 2.7
enstaller version: 4.6.1
sys.prefix: /Users/maye/Library/Enthought/Canopy_64bit/User
platform: Darwin-12.4.0-x86_64-i386-64bit
architecture: 64bit
use_webservice: True
config file: /Users/maye/.enstaller4rc
settings:
    prefix = /Users/maye/Library/Enthought/Canopy_64bit/User
    local = '/Users/maye/Library/Enthought/Canopy_64bit/User/LOCAL-REPO'
    noapp = False
    proxy = None
    IndexedRepos: (not used)
        'https://www.enthought.com/repo/epd/GPL-eggs/MacOSX/amd64/'
        'https://www.enthought.com/repo/epd/eggs/MacOSX/amd64/'
        'http://www.enthought.com/repo/pypi/eggs/MacOSX/amd64/'
You are logged in as [hidden email] (Michael Aye).
Subscription level: EPD Basic or above
()[maye@lunatic ~]$ enpkg --log
2013-02-15 17:04:46 -0800 PST  (rev 0)

2013-02-15 17:04:48 -0800 PST  (rev 1)
    +nose-1.2.1-1

2013-02-15 17:04:54 -0800 PST  (rev 2)
    +pandas-0.10.1-1

2013-02-15 17:05:00 -0800 PST  (rev 3)
    +swig-2.0.4-1

2013-02-15 17:05:12 -0800 PST  (rev 4)
    +sympy-0.7.2-1

2013-02-15 17:07:05 -0800 PST  (rev 5)
    +casuarius-1.0-1
    +enaml-0.2.0-1
    +traits-4.2.0-1

2013-02-15 17:07:27 -0800 PST  (rev 6)
    +enstaller-4.5.6-1

2013-02-15 17:09:10 -0800 PST  (rev 7)
    +opencv-2.4.2-1

2013-02-15 17:09:34 -0800 PST  (rev 8)
    +pyephem-3.7.5.1-1

2013-02-15 17:09:50 -0800 PST  (rev 9)
    +pymc-2.1b0-3

2013-02-19 17:15:51 -0800 PST  (rev 10)
    -enaml-0.2.0-1

2013-04-10 11:28:07 -0700 PDT  (rev 11)
    +apptools-4.2.0-1

2013-04-10 11:28:11 -0700 PDT  (rev 12)
    +blockcanvas-4.0.3-1

2013-04-10 11:28:15 -0700 PDT  (rev 13)
     casuarius  (1.0-1 -> 1.1-1)

2013-04-10 11:28:25 -0700 PDT  (rev 14)
    +chaco-4.3.0-1
    +enable-4.3.0-2

2013-04-10 11:28:27 -0700 PDT  (rev 15)
    +codetools-4.1.0-1

2013-04-10 11:28:34 -0700 PDT  (rev 16)
     traits  (4.2.0-1 -> 4.3.0-1)
    +enaml-0.6.8-1

2013-04-10 11:28:35 -0700 PDT  (rev 17)
     enstaller  (4.5.6-1 -> 4.6.1-2)

2013-04-10 11:29:03 -0700 PDT  (rev 18)
    +encore-0.3-1
    +envisage-4.3.0-1
    +ets-4.3.0-2
    +etsdevtools-4.0.2-1
    +etsproxy-0.1.2-1
    +graphcanvas-4.0.2-1
    +mayavi-4.3.0-2
    +pyface-4.3.0-1
    +scimath-4.1.2-1
    +traitsui-4.3.0-1

2013-04-10 11:29:04 -0700 PDT  (rev 19)
    +pyflakes-0.5.0-1

2013-04-10 11:30:40 -0700 PDT  (rev 20)
    +hdf-4.2.6-1

2013-04-24 16:48:43 -0700 PDT  (rev 21)
    +scikits.learn-0.8-2

2013-04-24 16:49:10 -0700 PDT  (rev 22)
    -scikits.learn-0.8-2

2013-04-24 17:00:50 -0700 PDT  (rev 23)
    +scikits.statsmodels-0.3.1-1

2013-04-24 17:04:08 -0700 PDT  (rev 24)
    -pandas-0.10.1-1

2013-04-29 20:24:03 -0700 PDT  (rev 25)
    +numpy-1.6.1-5

2013-04-29 20:24:04 -0700 PDT  (rev 26)
    +pandas-0.10.1-1

2013-04-29 20:24:08 -0700 PDT  (rev 27)
    +scikit_learn-0.13.1-3
    +scipy-0.12.0-1

2013-04-29 20:25:52 -0700 PDT  (rev 28)
    -pandas-0.10.1-1

2013-05-06 11:18:02 -0700 PDT  (rev 29)
    +basemap-1.0.6-1

2013-05-06 11:18:49 -0700 PDT  (rev 30)
    +pandas-0.11.0-1

2013-05-09 11:17:21 -0700 PDT  (rev 31)
    +statsmodels-0.4.3-1

2013-05-31 09:08:53 -0700 PDT  (rev 32)
    +mdp-3.3-1

2013-05-31 09:08:55 -0700 PDT  (rev 33)
    +pyzmq-2.2.0-2

2013-05-31 09:08:58 -0700 PDT  (rev 34)
    +scikits.image-0.8.2-1

2013-06-18 11:43:06 -0700 PDT  (rev 35)
     basemap  (1.0.6-1 -> 1.0.6-3)
     numpy  (1.6.1-5 -> 1.7.1-2)
    +matplotlib-1.2.0-7

2013-06-18 11:43:54 -0700 PDT  (rev 36)
     chaco  (4.3.0-1 -> 4.3.0-2)
     enable  (4.3.0-2 -> 4.3.0-4)

2013-06-18 11:44:03 -0700 PDT  (rev 37)
     enaml  (0.6.8-1 -> 0.6.8-2)
     traits  (4.3.0-1 -> 4.3.0-2)

2013-06-18 11:44:06 -0700 PDT  (rev 38)
    +networkx-1.6-2

2013-06-18 11:44:11 -0700 PDT  (rev 39)
    +pyproj-1.9.3-1

2013-06-18 11:44:15 -0700 PDT  (rev 40)
    +numexpr-2.0.1-3
    +pytables-2.4.0-1

2013-06-18 11:44:34 -0700 PDT  (rev 41)
     scipy  (0.12.0-1 -> 0.12.0-2)

2013-06-18 14:00:03 -0700 PDT  (rev 42)
    +biopython-1.59-2

2013-06-18 14:00:09 -0700 PDT  (rev 43)
    +fwrap-0.1.1-4

2013-06-18 14:00:20 -0700 PDT  (rev 44)
    +gdal-1.9.0-3
    +libgdal-1.9.0-2

2013-06-18 14:00:26 -0700 PDT  (rev 45)
    +h5py-2.1.3-2

2013-06-18 14:00:31 -0700 PDT  (rev 46)
    +lib_netcdf4-4.3.0-1

2013-06-18 14:00:35 -0700 PDT  (rev 47)
    +libxml2-2.7.8-3

2013-06-18 14:00:40 -0700 PDT  (rev 48)
    +libxslt-1.1.26-3

2013-06-18 14:00:44 -0700 PDT  (rev 49)
    +lxml-2.3.4-4

2013-06-18 14:00:56 -0700 PDT  (rev 50)
     matplotlib  (1.2.0-7 -> 1.2.1-1)

2013-06-18 14:01:10 -0700 PDT  (rev 51)
     mayavi  (4.3.0-2 -> 4.3.0-3)

2013-06-18 14:01:21 -0700 PDT  (rev 52)
     mdp  (3.3-1 -> 3.3-2)

2013-06-18 14:01:24 -0700 PDT  (rev 53)
    +netcdf4-1.0-4

2013-06-18 14:01:33 -0700 PDT  (rev 54)
     pandas  (0.11.0-1 -> 0.11.0-2)

2013-06-18 14:01:35 -0700 PDT  (rev 55)
    +pycluster-1.50-5

2013-06-18 14:01:37 -0700 PDT  (rev 56)
    +pyfits-3.0.6-2

2013-06-18 14:01:39 -0700 PDT  (rev 57)
    +pygarrayimage-0.0.7-5

2013-06-18 14:01:42 -0700 PDT  (rev 58)
    +pygments-1.6.0-1

2013-06-18 14:01:45 -0700 PDT  (rev 59)
    +pyhdf-0.8.3-8

2013-06-18 14:01:49 -0700 PDT  (rev 60)
     pymc  (2.1b0-3 -> 2.2.0-2)

2013-06-18 14:01:53 -0700 PDT  (rev 61)
     scikit_learn  (0.13.1-3 -> 0.13.1-4)

2013-06-18 14:01:55 -0700 PDT  (rev 62)
     scikits.image  (0.8.2-1 -> 0.8.2-2)

2013-06-18 14:01:58 -0700 PDT  (rev 63)
    +scikits.timeseries-0.91.3-5

2013-06-18 14:02:00 -0700 PDT  (rev 64)
     scimath  (4.1.2-1 -> 4.1.2-2)

2013-06-18 14:02:06 -0700 PDT  (rev 65)
     matplotlib  (1.2.1-1 -> 1.2.0-7)
    +shapely-1.2.17-1

2013-06-18 14:02:07 -0700 PDT  (rev 66)
    +simpy-2.2-2

2013-06-18 14:04:52 -0700 PDT  (rev 67)
     matplotlib  (1.2.0-7 -> 1.2.1-1)

2013-06-18 14:06:03 -0700 PDT  (rev 68)
    +nltk-2.0.1-2

2013-06-18 14:06:56 -0700 PDT  (rev 69)
    +pyodbc-3.0.6-1




_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

Michael Aye
In reply to this post by Jonathan March
>
> > 3) It seems that you did not try this, but it is worth noting that the
> > "Install All Canopy Packages" button is subject to the same issues as
> > "Update All..", but at a larger scale.
>
> I would like to do that but that list needs a clean up, there are 2
> overlapping packages in there, scikit_learn 0.13.1 and scikits.learn
> 0.8. Not sure if there are more.
>
> Yes, we are aware of this one. It's a little tricky because the old
> package needs to remain in the repo for legacy users (the same repo
> supports EPD), but we definitely should be excluding it in Canopy.
>
>  
There's another funny thing: basemap_ld requires mpl 1.2.0 it seems.
Because when I installed it, it removed mpl 1.2.1 and installed mpl
1.2.0. Promptly, the update manager tells me about mpl 1.2.1 being
available, which I then can install without the basemap_ld being
removed? So I guess I know have inconsistent packages installed. :(



_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev
Reply | Threaded
Open this post in threaded view
|

Re: Canopy status / updates / courses at Scipy

David Cournapeau


On Thu, Jun 20, 2013 at 1:35 PM, K.-Michael Aye <[hidden email]> wrote:
>
> > 3) It seems that you did not try this, but it is worth noting that the
> > "Install All Canopy Packages" button is subject to the same issues as
> > "Update All..", but at a larger scale.
>
> I would like to do that but that list needs a clean up, there are 2
> overlapping packages in there, scikit_learn 0.13.1 and scikits.learn
> 0.8. Not sure if there are more.
>
> Yes, we are aware of this one. It's a little tricky because the old
> package needs to remain in the repo for legacy users (the same repo
> supports EPD), but we definitely should be excluding it in Canopy.
>
>  
There's another funny thing: basemap_ld requires mpl 1.2.0 it seems.
Because when I installed it, it removed mpl 1.2.1 and installed mpl
1.2.0. Promptly, the update manager tells me about mpl 1.2.1 being
available, which I then can install without the basemap_ld being
removed? So I guess I know have inconsistent packages installed. :(

Michael, this is an oversight when I uploaded the new matplotlib 1.2.1. I have a fixed basemap egg that should fix the issue, and it should be available later today,

regards,
David

_______________________________________________
Enthought-Dev mailing list
[hidden email]
https://mail.enthought.com/mailman/listinfo/enthought-dev