OS X application bundle

0 views
Skip to first unread message

Maximilian Nickel

unread,
Jul 11, 2009, 11:56:06 AM7/11/09
to sage-devel
Hi,
i've just created a simple script that creates an OS X application
bundle for Sage. The usage is pretty simple, just copy the sage
directory from the .dmg into the directory of the script and run 'make
release'.
I thought it might be usefull for people running on OS X, so i'm
sharing it with this list.

You can get the code here: http://bitbucket.org/mnick/sageapp/

Best regards,
Max

gsw

unread,
Jul 13, 2009, 2:49:53 PM7/13/09
to sage-devel
Hi Max,
since everybody else seems to be too busy, I'll answer.

First of all, thanks for your effort, and thanks for sharing it with
us!

To be honest, the OS X fraction of the Sage Development Team is
currently rather inactive, e.g. one of us (Michael Abshoff) is taking
a temporary timeout from Sage development. Although there have been
two or three 99% attempts to finally deliver officially Sage on OS X
as an "app", this has not happened yet. IIRC, there even is currently
some switch via an environmental variable, that already does this, but
it is not on by default. As far as I remember, the problems still open
were with moving the app around, or copying it to a different computer
(which in the Apple world is a perfectly valid thing to do). Then Sage
upon startup should recognize this, rebuilding certain parts that
hardcode the path, or complaining if it was built e.g. on a PPC Mac
and now is copied to/started on an Intel Mac, or printing some
reasonable eror message if it was built under OS X 10.5 and/or with
"64bit" enabled, and now is started under OS X 10.4.

Personally, I hope that "OS X 64bi"t will be added to the released
binaries in the Sage 4.1.1 version. But I fear that the long overdue
"cleanup" of certain oddities like the outdated "Readme-OS-X", or the
fact that in the .dmg, Sage lacks a verison number (it should be e.g a
folder "sage-4.1" instead of just "sage") will not happen before OS X
10.6 is out and triggers this.

Cheers,
Georg

William Stein

unread,
Jul 13, 2009, 2:58:47 PM7/13/09
to sage-devel
2009/7/13 gsw <Georg...@googlemail.com>:

>
> Hi Max,
>
> On 11 Jul., 17:56, Maximilian Nickel <max.nic...@gmail.com> wrote:
>> Hi,
>> i've just created a simple script that creates an OS X application
>> bundle for Sage. The usage is pretty simple, just copy the sage
>> directory from the .dmg into the directory of the script and run 'make
>> release'.
>> I thought it might be usefull for people running on OS X, so i'm
>> sharing it with this list.
>>
>> You can get the code here:http://bitbucket.org/mnick/sageapp/
>>
>> Best regards,
>> Max
>
> since everybody else seems to be too busy, I'll answer.
>
> First of all, thanks for your effort, and thanks for sharing it with
> us!
>
> To be honest, the OS X fraction of the Sage Development Team is
> currently rather inactive, e.g. one of us (Michael Abshoff) is taking
> a temporary timeout from Sage development.

Michael Abshoff is no longer involved in Sage development. Almost
the entire time he worked on Sage development he did so as a fulltime
paid employee and the source of funding that paid for him has ran out.
He is of course 100% welcome back, but if he comes back it will be
as a volunteer like most everybody else.

That said, a huge proportion (maybe half?) of the Sage developers use
OS X, and getting an app bundle version of Sage is very welcome.
Last year and the year before I devoted substantial time to trying to
do this. I gave up in frustration. Then Michael A. tried to
integrate in something from somebody else, but also gave up in
frustration. This pattern has been repeated several times.

> Although there have been
> two or three 99% attempts to finally deliver officially Sage on OS X
> as an "app", this has not happened yet. IIRC, there even is currently
> some switch via an environmental variable, that already does this, but
> it is not on by default. As far as I remember, the problems still open
> were with moving the app around, or copying it to a different computer
> (which in the Apple world is a perfectly valid thing to do). Then Sage
> upon startup should recognize this, rebuilding certain parts that
> hardcode the path, or complaining if it was built e.g. on a PPC Mac
> and now is copied to/started on an Intel Mac, or printing some
> reasonable eror message if it was built under OS X 10.5 and/or with
> "64bit" enabled, and now is started under OS X 10.4.
>
> Personally, I hope that "OS X 64bi"t will be added to the released
> binaries in the Sage 4.1.1 version.

I posted a binary already for OS X 64-bit:

http://sage.math.washington.edu/home/wstein/binaries/

Minh or Harald should hopefully post that to the sage.mathorg binaries
page (hint hint).
It's really easy to build OS X 64-bit.

> But I fear that the long overdue
> "cleanup" of certain oddities like the outdated "Readme-OS-X", or the
> fact that in the .dmg, Sage lacks a verison number (it should be e.g a
> folder "sage-4.1" instead of just "sage") will not happen before OS X
> 10.6 is out and triggers this.

Or... maybe Maximilian Nickel will help with some of that.

I think most

Minh Nguyen

unread,
Jul 13, 2009, 3:03:43 PM7/13/09
to sage-...@googlegroups.com
On Tue, Jul 14, 2009 at 4:49 AM, gsw<Georg...@googlemail.com> wrote:

<SNIP>

> To be honest, the OS X fraction of the Sage Development Team is
> currently rather inactive, e.g. one of us (Michael Abshoff) is taking
> a temporary timeout from Sage development. Although there have been
> two or three 99% attempts to finally deliver officially Sage on OS X
> as an "app", this has not happened yet. IIRC, there even is currently
> some switch via an environmental variable, that already does this, but
> it is not on by default.

The following post documents how to build an app on Mac OS X:

http://mvngu.wordpress.com/2009/03/22/clickable-mac-os-x-app-for-sage-34/

As you can tell, it's a shameless act of self-referral :-)

The following thread is related to this discussion:

http://groups.google.com/group/sage-support/browse_thread/thread/623e6a64bafcb7ff?pli=1
--
Regards
Minh Van Nguyen

Minh Nguyen

unread,
Jul 13, 2009, 3:10:09 PM7/13/09
to sage-...@googlegroups.com
On Tue, Jul 14, 2009 at 4:58 AM, William Stein<wst...@gmail.com> wrote:

<SNIP>

> I posted a binary already for OS X 64-bit:
>
> http://sage.math.washington.edu/home/wstein/binaries/
>
> Minh or Harald should hopefully post that to the sage.mathorg binaries
> page (hint hint).
> It's really easy to build OS X 64-bit.

Those OS X binaries are already up on the master mirror. All binaries
have been mirrored out to servers around the world. The only mirrors
that are lagging behind everyone else are the Australian mirrors (you
don't call Australia as the land from down under for nothing :-)

We have OS X binaries for Intel as well as PowerPC, including 32- and
64-bit versions, and for OS X 10.4 and 10.5. The 10.4 binaries are due
to Georg S. Weber who has been building them for many releases now.
Thanks, Georg!

Maximilian Nickel

unread,
Jul 13, 2009, 3:23:41 PM7/13/09
to sage-...@googlegroups.com
Hi Georg,
thanks for your answers, i didn't know there was already code to generate an application bundle in sage. I'll take a look at it
 
As far as I remember, the problems still open
were with moving the app around, or copying it to a different computer
(which in the Apple world is a perfectly valid thing to do). Then Sage
upon startup should recognize this, rebuilding certain parts that
hardcode the path, or complaining if it was built e.g. on a PPC Mac
and now is copied to/started on an Intel Mac, or printing some
reasonable eror message if it was built under OS X 10.5 and/or with
"64bit" enabled, and now is started under OS X 10.4.

Moving the Sage.app around on the same computer seems to work ok, or at least
the python files that hardcode the path get regenerated.
I'll see if i can come up with something about the OS version/arch issue
  
Cheers
Max

Reply all
Reply to author
Forward
0 new messages