OS X Mavericks

1823 views
Skip to first unread message

Jason Grout

unread,
Oct 22, 2013, 1:30:47 PM10/22/13
to sage-...@googlegroups.com
Apple just announced OS X Mavericks is free, available today. Has
anyone tried compiling Sage on it?

Thanks,

Jason

William Stein

unread,
Oct 22, 2013, 1:42:08 PM10/22/13
to sage-devel
On Tue, Oct 22, 2013 at 10:30 AM, Jason Grout
<jason...@creativetrax.com> wrote:
> Apple just announced OS X Mavericks is free, available today. Has anyone
> tried compiling Sage on it?

Does that mean I can install it into a virtualbox (say) machine on my
OS X 10.6 desktop (bsd.math), so people can try it... but we still
have a 10.6 buildbot?

William

>
> Thanks,
>
> Jason
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/groups/opt_out.



--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

yomcat

unread,
Oct 22, 2013, 4:29:35 PM10/22/13
to sage-...@googlegroups.com


On Wednesday, 23 October 2013 06:30:47 UTC+13, jason wrote:
Apple just announced OS X Mavericks is free, available today.  Has
anyone tried compiling Sage on it?

I have just installed it. I shall build Sage and report back. One thing I noticed is that XCode has done something different with the command-line tools. I still seem to have them, but that might be because I already had them when I upgraded.

yomcat

unread,
Oct 22, 2013, 4:58:49 PM10/22/13
to sage-...@googlegroups.com

The end was:

The directory that should contain system headers does not exist:

  /usr/include

make[6]: *** [stmp-fixinc] Error 1

make[5]: *** [all-stage1-gcc] Error 2

make[4]: *** [stage1-bubble] Error 2

make[3]: *** [all] Error 2


real 4m32.385s

user 3m23.834s

sys 1m0.647s

************************************************************************

Error installing package gcc-4.7.3.p0

************************************************************************

Please email sage-devel (http://groups.google.com/group/sage-devel)

explaining the problem and including the relevant part of the log file

  /Users/yomcat/Downloads/sage-test/logs/pkgs/gcc-4.7.3.p0.log

Describe your computer, operating system, etc.

If you want to try to fix the problem yourself, *don't* just cd to

/Users/yomcat/Downloads/sage-test/spkg/build/gcc-4.7.3.p0 and type 'make' or whatever is appropriate.

Instead, the following commands setup all environment variables

correctly and load a subshell for you to debug the error:

  (cd '/Users/yomcat/Downloads/sage-test/spkg/build/gcc-4.7.3.p0' && '/Users/yomcat/Downloads/sage-test/sage' --sh)

When you are done debugging, you can type "exit" to leave the subshell.

************************************************************************

make[2]: *** [/Users/yomcat/Downloads/sage-test/spkg/installed/gcc-4.7.3.p0] Error 1

make[1]: *** [all] Error 2


There seems to be an include directory a level down (in llvm-gcc-4.2). I've sym-linked that in, we'll see what happens now. 

yomcat

unread,
Oct 22, 2013, 5:04:57 PM10/22/13
to sage-...@googlegroups.com


On Wednesday, 23 October 2013 09:58:49 UTC+13, yomcat wrote:

There seems to be an include directory a level down (in llvm-gcc-4.2). I've sym-linked that in, we'll see what happens now. 

It still broke, but further on. 

/Users/yomcat/Downloads/sage-test/spkg/build/gcc-4.7.3.p0/gcc-build/./gcc/xgcc -B/Users/yomcat/Downloads/sage-test/spkg/build/gcc-4.7.3.p0/gcc-build/./gcc/ -B/Users/yomcat/Downloads/sage-test/local/x86_64-apple-darwin13.0.0/bin/ -B/Users/yomcat/Downloads/sage-test/local/x86_64-apple-darwin13.0.0/lib/ -isystem /Users/yomcat/Downloads/sage-test/local/x86_64-apple-darwin13.0.0/include -isystem /Users/yomcat/Downloads/sage-test/local/x86_64-apple-darwin13.0.0/sys-include    -g -O2 -O2  -g -O2 -DIN_GCC   -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition  -isystem ./include   -pipe -fno-common -g -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector   -pipe -fno-common -I. -I. -I../.././gcc -I../../../src/libgcc -I../../../src/libgcc/. -I../../../src/libgcc/../gcc -I../../../src/libgcc/../include  -DHAVE_CC_TLS -DUSE_EMUTLS -o _muldi3_s.o -MT _muldi3_s.o -MD -MP -MF _muldi3_s.dep -DSHARED -DL_muldi3 -c ../../../src/libgcc/libgcc2.c

In file included from ../../../src/libgcc/libgcc2.c:29:0:

../../../src/libgcc/../gcc/tsystem.h:88:19: fatal error: stdio.h: No such file or directory

compilation terminated.

make[6]: *** [_muldi3_s.o] Error 1

make[5]: *** [all-stage1-target-libgcc] Error 2

make[4]: *** [stage1-bubble] Error 2

make[3]: *** [all] Error 2

I'll go hunting for dev tools. 

yomcat

unread,
Oct 22, 2013, 5:48:00 PM10/22/13
to sage-...@googlegroups.com


On Wednesday, 23 October 2013 10:04:57 UTC+13, yomcat wrote:

I'll go hunting for dev tools. 

They're not out yet.

yomcat

unread,
Oct 22, 2013, 8:48:24 PM10/22/13
to sage-...@googlegroups.com
Got some. The build got much further, but broke on polybori - https://dl.dropboxusercontent.com/u/32868672/polybori-0.8.3.log 

Jason Grout

unread,
Oct 22, 2013, 8:52:07 PM10/22/13
to sage-...@googlegroups.com
On 10/22/13 12:42 PM, William Stein wrote:
> Does that mean I can install it into a virtualbox (say) machine on my
> OS X 10.6 desktop (bsd.math), so people can try it... but we still
> have a 10.6 buildbot?

It looks like you probably can install it in a virtual machine. I have
no clue how easy it is.

Jason


William Stein

unread,
Oct 22, 2013, 9:17:25 PM10/22/13
to sage-devel
On Tue, Oct 22, 2013 at 1:29 PM, yomcat <yom...@gmail.com> wrote:
>
>
> On Wednesday, 23 October 2013 06:30:47 UTC+13, jason wrote:
>>
>> Apple just announced OS X Mavericks is free, available today. Has
>> anyone tried compiling Sage on it?
>
>
> I have just installed it. I shall build Sage and report back. One thing I

Can you also try installing our OS X 10.8 binary on it, and seeing if
it works at all?

William

> noticed is that XCode has done something different with the command-line
> tools. I still seem to have them, but that might be because I already had
> them when I upgraded.
>

yomcat

unread,
Oct 22, 2013, 9:49:20 PM10/22/13
to sage-...@googlegroups.com


On Wednesday, 23 October 2013 14:17:25 UTC+13, William wrote:
On Tue, Oct 22, 2013 at 1:29 PM, yomcat <yom...@gmail.com> wrote:
>
>
> On Wednesday, 23 October 2013 06:30:47 UTC+13, jason wrote:
>>
>> Apple just announced OS X Mavericks is free, available today.  Has
>> anyone tried compiling Sage on it?
>
>
> I have just installed it. I shall build Sage and report back. One thing I

Can you also try installing our OS X 10.8 binary on it, and seeing if
it works at all?

No problems at all. I didn't try the -app version, but can do if requested. 

William

William Stein

unread,
Oct 22, 2013, 9:57:38 PM10/22/13
to sage-devel
That's great news. Could you try running the test suite, i.e., go to
root folder of the install and type

make test

I don't know if enough dev tools are there to run make though... If
not, you could do

./sage -t devel/sage/sage

which is more or less the same.

William
>>
>>
>> William

yomcat

unread,
Oct 22, 2013, 11:36:12 PM10/22/13
to sage-...@googlegroups.com


On Wednesday, 23 October 2013 14:57:38 UTC+13, William wrote:
On Tue, Oct 22, 2013 at 6:49 PM, yomcat <yom...@gmail.com> wrote:
>
>
> On Wednesday, 23 October 2013 14:17:25 UTC+13, William wrote:
>>
>> On Tue, Oct 22, 2013 at 1:29 PM, yomcat <yom...@gmail.com> wrote:
>> >
>> >
>> > On Wednesday, 23 October 2013 06:30:47 UTC+13, jason wrote:
>> >>
>> >> Apple just announced OS X Mavericks is free, available today.  Has
>> >> anyone tried compiling Sage on it?
>> >
>> >
>> > I have just installed it. I shall build Sage and report back. One thing
>> > I
>>
>> Can you also try installing our OS X 10.8 binary on it, and seeing if
>> it works at all?
>
>
> No problems at all. I didn't try the -app version, but can do if requested.

That's great news.  Could you try running the test suite, i.e., go to
root folder of the install and type

     make test

It fails at various points. I have to go home now, but the log is at https://dl.dropboxusercontent.com/u/32868672/test.log 

Jeroen Demeyer

unread,
Oct 23, 2013, 2:36:14 AM10/23/13
to sage-...@googlegroups.com
Can you build Sage with MAKE="make -k"? This will force the build to
continue as much as possible, even if some package fails. Please report
which other packages (if any) fail to build.

John H Palmieri

unread,
Oct 23, 2013, 2:01:25 PM10/23/13
to sage-...@googlegroups.com

Here's what I would suggest:

- first, there is a new version of Xcode (5.0.1, released Oct 22), so install that.
- then run Xcode so it installs whatever it thinks it needs to install.
- You might also need to manually install the command-line tools: from the terminal, run

  xcode-select --install

and click "Install" in the pop-up. (I don't know if this step is required, but I did it.)

Then try building Sage. I'm trying this right now, but I've only installed OS X Mavericks on an old slow machine, so it will be a while before I know if it works.

--
John

Michael Welsh

unread,
Oct 23, 2013, 2:08:50 PM10/23/13
to sage-...@googlegroups.com

On 24/10/2013, at 0701, John H Palmieri <jhpalm...@gmail.com> wrote:

> - You might also need to manually install the command-line tools: from the terminal, run
>
> xcode-select --install
>
> and click "Install" in the pop-up. (I don't know if this step is required, but I did it.)

This step is required, but I just grabbed the tools from ADC. Your way is "easier" - maybe it should be in the readme (especially now that CLT have disappeared as a download from inside XCode, with a promise to "download on the fly the first time they're required")

I'm building it now on my fast machine. Serial builds are so slow...

John H Palmieri

unread,
Oct 23, 2013, 3:23:08 PM10/23/13
to sage-...@googlegroups.com


On Wednesday, October 23, 2013 11:08:50 AM UTC-7, yomcat wrote:

On 24/10/2013, at 0701, John H Palmieri <jhpalm...@gmail.com> wrote:

> - You might also need to manually install the command-line tools: from the terminal, run
>
>   xcode-select --install
>
> and click "Install" in the pop-up. (I don't know if this step is required, but I did it.)

This step is required, but I just grabbed the tools from ADC. Your way is "easier" - maybe it should be in the readme

Michael Welsh

unread,
Oct 23, 2013, 4:06:33 PM10/23/13
to sage-...@googlegroups.com

On 23/10/2013, at 1936, Jeroen Demeyer <jdem...@cage.ugent.be> wrote:

> Can you build Sage with MAKE="make -k"? This will force the build to continue as much as possible, even if some package fails. Please report which other packages (if any) fail to build.

It's finished:

The following package(s) may have failed to build:

package: polybori-0.8.3
log file: /Users/yomcat/Desktop/sage-test/logs/pkgs/polybori-0.8.3.log
build directory: /Users/yomcat/Desktop/sage-test/spkg/build/polybori-0.8.3

package: r-2.15.2.p2
log file: /Users/yomcat/Desktop/sage-test/logs/pkgs/r-2.15.2.p2.log
build directory: /Users/yomcat/Desktop/sage-test/spkg/build/r-2.15.2.p2

package: scipy-0.12.0.p0
log file: /Users/yomcat/Desktop/sage-test/logs/pkgs/scipy-0.12.0.p0.log
build directory: /Users/yomcat/Desktop/sage-test/spkg/build/scipy-0.12.0.p0

package: singular-3-1-5.p9
log file: /Users/yomcat/Desktop/sage-test/logs/pkgs/singular-3-1-5.p9.log
build directory: /Users/yomcat/Desktop/sage-test/spkg/build/singular-3-1-5.p9

All the logs are at https://www.dropbox.com/sh/pmzh03082qe4nu1/OGfJ0z9H6W/logs

Michael

Ben Salisbury

unread,
Oct 24, 2013, 2:50:20 PM10/24/13
to sage-...@googlegroups.com
I am also having an issue building on OS X 10.9, though (I admit) I am trying to build the 5.13beta1 version.  My build fails at the polybori step:

Error installing PolyBoRi.


real 1m14.940s

user 1m8.348s

sys 0m4.921s

************************************************************************

Error installing package polybori-0.8.3

************************************************************************


Log file is available at https://www.dropbox.com/s/ift6cmsilov8sqd/polybori-0.8.3.log.  

Any solutions or suggestions?  (I'm making with 'make -k' now.)

Ben

yomcat

unread,
Oct 24, 2013, 2:54:03 PM10/24/13
to sage-...@googlegroups.com
The full copy/paste from Terminal is at https://dl.dropboxusercontent.com/u/32868672/test.txt

Michael 

Chau Nguyen

unread,
Oct 28, 2013, 4:37:18 AM10/28/13
to sage-...@googlegroups.com
I had the same problem when building sage 5.12. I used simple "make".

My log files are

https://dl.dropboxusercontent.com/u/68207449/install.log

https://dl.dropboxusercontent.com/u/68207449/polybori-0.8.3.log

Best regards,
Chau.

Dima Pasechnik

unread,
Oct 28, 2013, 5:59:47 AM10/28/13
to sage-...@googlegroups.com
On 2013-10-28, Chau Nguyen <chaungu...@gmail.com> wrote:
> On Thursday, October 24, 2013 8:50:20 PM UTC+2, Ben Salisbury wrote:
>> I am also having an issue building on OS X 10.9, though (I admit) I am trying to build the 5.13beta1 version.  My build fails at the polybori step:
>>
>>
Install file:
"build/Users/chaunguyen/Installs/sage-5.12/local/lib/libpolybori-0.8.dylib.3.0.0"
as
"/Users/chaunguyen/Installs/sage-5.12/local/lib/libpolybori-0.8.dylib.3.0.0"
Chmod("/Users/chaunguyen/Installs/sage-5.12/local/lib/libpolybori-0.8.dylib.3.0.0",
0644)
install_name_tool -id @loader_path/libpolybori-0.8.dylib.3.0.0
/Users/chaunguyen/Installs/sage-5.12/local/lib/libpolybori-0.8.dylib.3.0.0
sh: line 1: 98559 Trace/BPT trap: 5
/Applications/Xcode.app/Contents/Developer/usr/bin/xcodebuild -sdk /
-find install_name_tool 2> /dev/null
install_name_tool: error: unable to find utility "install_name_tool",
not a developer tool or in PATH
scons: ***
[/Users/chaunguyen/Installs/sage-5.12/local/lib/libpolybori-0.8.dylib.3.0.0]
Error 72
scons: building terminated because of errors.
Error installing PolyBoRi.


it seems that install_name_tool has been dropped in XCode 5, hence these
errors in polybory.
E.g. on OSX 10.6 I have it:

$ which install_name_tool
/usr/bin/install_name_tool

Do you have it on your system?

Chau Nguyen

unread,
Oct 28, 2013, 6:50:54 AM10/28/13
to sage-...@googlegroups.com
Yes, I actually do have install_name_tool as well

$ which install_name_tool
/usr/bin/install_name_tool

Which is also in PATH

$ echo $PATH
/opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/opt/X11/bin:/usr/texbin

Maybe I try rebuild in the after noon and let you know more... (it took some 3 hours to reach the error)

Thanks,
Chau.
> --
> You received this message because you are subscribed to a topic in the Google Groups "sage-devel" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/sage-devel/5ne3SwEeYAE/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to sage-devel+...@googlegroups.com.

Dima Pasechnik

unread,
Oct 28, 2013, 7:11:48 AM10/28/13
to sage-...@googlegroups.com
On 2013-10-28, Chau Nguyen <chaungu...@gmail.com> wrote:
> Yes, I actually do have install_name_tool as well
>
> $ which install_name_tool
> /usr/bin/install_name_tool
>
> Which is also in PATH

hmm, weird, can you run the following line in the terminal:
$ xcodebuild -sdk / -find install_name_tool

for me (on OSX 10.6) it works just fine and prints
/Developer/usr/bin/install_name_tool



> $ echo $PATH
> /opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/opt/X11/bin:/usr/texbin
>
> Maybe I try rebuild in the after noon and let you know more... (it took some 3 hours to reach the error)
>

Could it be that you have some stuff in /opt/local/bin or/and /opt/local/sbin that affects
the build?

It could be that there is something that takes precedence over the
"standard" things in /usr/bin:/bin:/usr/sbin:/sbin

So the next thing to try would be to remove
/opt/local/bin:/opt/local/sbin: from your PATH and try building from the
start (after "make distclean" or "make clean") again.

Chau Nguyen

unread,
Oct 28, 2013, 9:14:19 AM10/28/13
to sage-...@googlegroups.com
Yes, the command works for me as well

$ xcodebuild -sdk / -find install_name_tool
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/install_name_tool

I also renamed "opt/" directory before building...

Anyway, I am building again and will inform you about this when it finishes. ("make clean" did not help: the same error happens after a short time, I had to run "make distclean")

Thanks a lot for your helps!
Chau.

Jeroen Demeyer

unread,
Oct 28, 2013, 10:09:56 AM10/28/13
to sage-...@googlegroups.com
Thanks for the logs. Unfortunately, I see no obvious fix for any of
these errors :-(

Jeroen.

Chau Nguyen

unread,
Oct 28, 2013, 12:03:56 PM10/28/13
to sage-...@googlegroups.com
No it still stops with the same error.

Perhaps I figured out what is wrong with my case (sorry, maybe I should not have mixed with the existing discussion of Jeroen Demeyer).

The script tried to run

/Applications/Xcode.app/Contents/Developer/usr/bin/xcodebuild -sdk / -find install_name_tool 2> /dev/null
install_name_tool: error: unable to find utility "install_name_tool", not a developer tool or in PATH

The install_name_tool is in fact in an other place (Xcode 5.0.1)

$ xcodebuild -sdk / -find install_name_tool
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/install_name_tool

But I do not know how to fix this. Do you have any suggestion? Thank you!

Regards,
Chau.

Dima Pasechnik

unread,
Oct 28, 2013, 12:21:56 PM10/28/13
to sage-...@googlegroups.com
On 2013-10-28, Chau Nguyen <chaungu...@gmail.com> wrote:
> No it still stops with the same error.
>
> Perhaps I figured out what is wrong with my case (sorry, maybe I should not have mixed with the existing discussion of Jeroen Demeyer).
>
> The script tried to run
>
> /Applications/Xcode.app/Contents/Developer/usr/bin/xcodebuild -sdk / -find install_name_tool 2> /dev/null
> install_name_tool: error: unable to find utility "install_name_tool", not a developer tool or in PATH
>
> The install_name_tool is in fact in an other place (Xcode 5.0.1)
>
> $ xcodebuild -sdk / -find install_name_tool
> /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/install_name_tool
>
> But I do not know how to fix this. Do you have any suggestion? Thank you!

What happens if you run

/Applications/Xcode.app/Contents/Developer/usr/bin/xcodebuild -sdk / -find install_name_tool

at the command prompt?
(It could be that you have different incarnations of xcodebuild in your
path...)

Chau Nguyen

unread,
Oct 28, 2013, 12:33:29 PM10/28/13
to sage-...@googlegroups.com
Yes, you are right, it still works (I think)

$/Applications/Xcode.app/Contents/Developer/usr/bin/xcodebuild -sdk / -find install_name_tool
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/install_name_tool


Chau.

Chau Nguyen

unread,
Oct 28, 2013, 1:10:30 PM10/28/13
to sage-...@googlegroups.com
I tried running make -k, and now I am at the same errors as Michael Welsh...

Best regards,
Chau.

Mike Zabrocki

unread,
Nov 4, 2013, 8:56:50 PM11/4/13
to sage-...@googlegroups.com
I had updated my Xcode and OS system to 10.9 Mavericks in late October.  I was having no problems building and working with patches for a long time until I touched some files that required certain certain libraries.  When I touched those files I was no longer able to complete 'sage -b'.  

When I tried downloading a copy from source and compiling I was getting only about 5-10 minutes into the compile before I was getting an error.  I was getting an error long before getting to the polybori stage.

It turns out I needed to download a new copy of the command line tools.  To do this I completed the following:
1. Open Xcode
2. from the Xcode menu -> Open Developer Tool -> More Developer Tools ...
3. I needed to log into the Apple Developer website
4. There was a command line tools for OSX Mavericks which I downloaded and installed
Note that there is also an update for OSX Mountain Lion so this should not only be the problem for 10.9

After I installed the new command line tools then I was able to build from source also for about 3 hours and then I encountered the same errors that others are seeing here.

-Mike

William Stein

unread,
Nov 6, 2013, 5:20:00 PM11/6/13
to sage-devel
On Mon, Nov 4, 2013 at 5:56 PM, Mike Zabrocki <mike.z...@gmail.com> wrote:
>[...]
> After I installed the new command line tools then I was able to build from
> source also for about 3 hours and then I encountered the same errors that
> others are seeing here.

Hi,

In the interest of porting Sage to OS X 10.9, I dug up a macbook pro
laptop, did a clean install of OS X 10.9 and dev tools on it, and put
it on the web. I've made project on https://cloud.sagemath.com with
ssh key access to an account on that laptop. Anybody who is
interested, send me an email (wst...@gmail.com) and I'll add you to
the project, so we can work on porting together.

(I can also just send you a private key so you can connect to the
laptop via ssh from wherever.)

-- William

Jean-Pierre Flori

unread,
Nov 6, 2013, 5:34:25 PM11/6/13
to sage-...@googlegroups.com


On Wednesday, November 6, 2013 11:20:00 PM UTC+1, William wrote:
On Mon, Nov 4, 2013 at 5:56 PM, Mike Zabrocki <mike.z...@gmail.com> wrote:
>[...]
> After I installed the new command line tools then I was able to build from
> source also for about 3 hours and then I encountered the same errors that
> others are seeing here.

Hi,

In the interest of porting Sage to OS X 10.9, I dug up a macbook pro
laptop, did a clean install of OS X 10.9 and dev tools on it, and put
 
When I get a little more time, I'll try to install Mavericks into a virtual machine (if possible, Lion was quite easy) and give it a shot as well.

Best,
JP

William Stein

unread,
Nov 6, 2013, 5:39:38 PM11/6/13
to sage-...@googlegroups.com
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Well I just added you to the Sagemath Cloud project, in the hopes you
can't resist taking a look


https://cloud.sagemath.com/projects/4148c5ce-7a49-43fd-b558-359172738584/files/

> JP
>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.22 (Darwin)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=OzoW
-----END PGP SIGNATURE-----

R. Andrew Ohana

unread,
Nov 16, 2013, 12:02:53 AM11/16/13
to sage-...@googlegroups.com
Has anyone else been looking into this?

Where I've gotten:

scipy:
  easy fix, just add "-D__ACCELERATE__" to CPPFLAGS

polybori:
  this is a phantom bug that I can no longer reproduce (and I don't know what changed in my environment)

r:
  seems less like an issue with r, than with sqlite -- in particular it seems CoreData relies more heavily on sqlite now, and the version we build is missing some stuff that it needs (this is an unfortunate side-effect of setting DYLD_LIBRARY_PATH)

sage library:
  chokes on -latlas, although setting BLAS2="cblas" seems to resolve the issue

other:
  upon startup, the sage runtime crashes -- the copy of libstdc++ we build can't seem to find some stuff



--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To post to this group, send email to sage-...@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/groups/opt_out.



--
Andrew

Jeroen Demeyer

unread,
Nov 16, 2013, 3:20:36 AM11/16/13
to sage-...@googlegroups.com, William Stein
On 2013-11-16 06:02, R. Andrew Ohana wrote:
> Has anyone else been looking into this?
Yes, I tried William's laptop (which is down now). But I didn't get far,
since /usr/include is missing. Somebody needs to properly install XCode
on that machine before I can continue.

William Stein

unread,
Nov 16, 2013, 3:22:26 AM11/16/13
to Jeroen Demeyer, sage-devel
I spent a while clicking around, trying what you suggested, searching,
etc., with XCode, but no matter what I did /usr/include never got
installed. Also, yesterday, I stupidly didn't reconnect the
laptop to the correct network, which is why it isn't accessible right
now.

Jeroen Demeyer

unread,
Nov 16, 2013, 3:26:11 AM11/16/13
to sage-...@googlegroups.com
On 2013-10-23 02:48, yomcat wrote:
>
>
> On Wednesday, 23 October 2013 10:48:00 UTC+13, yomcat wrote:
>
>
>
> On Wednesday, 23 October 2013 10:04:57 UTC+13, yomcat wrote:
>
>
> I'll go hunting for dev tools.
>
>
> They're not out yet.
>
>
> Got some.
Perhaps you should explain the steps that you did to get this far with
building Sage, such that other people can continue from there.

Michael Welsh

unread,
Nov 16, 2013, 3:55:07 AM11/16/13
to sage-...@googlegroups.com
My method isn't easily reproducible.

Now, one can in theory install the dev tools by trying to run make. OS X should then ask you to download and install them (much like it does with X11). Alternatively (this is what will go in the readme), one can run the command xcode-select --install and click "install" in the resulting pop-up. (#15319). The other way is to log in to developer.apple.com and find the download.

John H Palmieri

unread,
Nov 16, 2013, 11:48:11 AM11/16/13
to sage-...@googlegroups.com


I installed the latest version of Xcode (via the App store) and then ran "xcode-select --install" from the terminal. All of the files in /usr/include have recent modification dates, so I think they were put there by this method.

Then I ran 'make -k' and got failures in polybori (still -- I tried again last night), r, and singular.

--
John

R. Andrew Ohana

unread,
Nov 17, 2013, 1:18:00 AM11/17/13
to sage-...@googlegroups.com
Ok, I've successfully built a fully working copy of sage on 10.9 (i.e. passes all
doctests with 3 exceptions, see below).

polybori: I stopped experiencing the issue others are having (and still have not
been able to figure out what resolved it for me)

scipy: This seems like more or less the same issue that we had with 10.8 -- see
#13309. An easy work around until scipy is updated is to add
-D__ACCELERATE__ to the CPPFLAGS.

r: links against CoreData, which in turn links against sqlite. Unfortunately, it
seems like CoreData now requires the extra modules Apple has included in the
system copy of sqlite. As I see it we have 2 options:
1) include these extra modules to support CoreData
2) switch to using the system sqlite on osx
I would vote for the 2nd, as I think it would reduce future headache (and it isn't
crucial for our purposes to have the most recent version of sqlite). Doing this
currently produces 3 doctests errors that explicitly test our copy of sqlite

sage library: the system cblas won't pretend it is atlas anymore, so set
BLAS2 to 'cblas'

gcc: at least for me, I was finding that the resulting libstdc++ was missing some
symbols. I haven't looked into this yet (I instead used homebrew's gcc).

--
Andrew

R. Andrew Ohana

unread,
Nov 18, 2013, 2:51:59 AM11/18/13
to sage-...@googlegroups.com
I've opened #15433 to trac porting to 10.9.
--
Andrew

R. Andrew Ohana

unread,
Nov 18, 2013, 10:07:42 PM11/18/13
to sage-...@googlegroups.com
I've uploaded a binary to http://boxen.math.washington.edu/home/ohanar/builds/sage-5.13.beta3-x86_64-Darwin.dmg -- no promises it will work for you.
--
Andrew

David Joyner

unread,
Nov 19, 2013, 12:09:31 AM11/19/13
to sage-devel
On Mon, Nov 18, 2013 at 10:07 PM, R. Andrew Ohana
<andrew...@gmail.com> wrote:
> I've uploaded a binary to
> http://boxen.math.washington.edu/home/ohanar/builds/sage-5.13.beta3-x86_64-Darwin.dmg
> -- no promises it will work for you.
>

This worked for me. Thank you!

William Stein

unread,
Nov 20, 2013, 4:59:38 PM11/20/13
to sage-devel
On Mon, Nov 18, 2013 at 7:07 PM, R. Andrew Ohana <andrew...@gmail.com> wrote:
> I've uploaded a binary to
> http://boxen.math.washington.edu/home/ohanar/builds/sage-5.13.beta3-x86_64-Darwin.dmg
> -- no promises it will work for you.
>

I just saw this, and barely looked at it, but it may be useful:

"OS X 10.9 Mavericks, Python, & Canopy"
https://support.enthought.com/entries/22861925-OS-X-10-9-Mavericks-Python-Canopy
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/groups/opt_out.



R. Andrew Ohana

unread,
Nov 20, 2013, 11:57:58 PM11/20/13
to sage-...@googlegroups.com
FYI, this is up for review now.
--
Andrew

Stefan

unread,
Nov 22, 2013, 10:47:29 PM11/22/13
to sage-...@googlegroups.com
Just wondering... has Apple's compiler by any chance improved enough so we can stop building our own GCC?

R. Andrew Ohana

unread,
Nov 23, 2013, 12:52:28 AM11/23/13
to sage-...@googlegroups.com

They don't supply a Fortran compiler, which we need. Also as of Xcode 5 they no longer supply gcc (the gcc in PATH is actually clang) and some packages we include depend explicitly on having gcc as our c compiler.

If anything we need our own compiler now more than before, and this situation doesn't look like it will be changing given Apple's decisions.

Andrew

On Nov 22, 2013 7:47 PM, "Stefan" <stefan...@gmail.com> wrote:
Just wondering... has Apple's compiler by any chance improved enough so we can stop building our own GCC?

--

Taekyung Kim

unread,
Nov 23, 2013, 3:21:46 AM11/23/13
to sage-...@googlegroups.com
Hi.

I have the same problem with r while building sage 5.12 on my mavericks macbook air.

You proposed to "switch to using the system sqlite on osx", but how can I do this?

Thanks.


Taekyung



2013년 11월 17일 일요일 오후 3시 18분 0초 UTC+9, R. Andrew Ohana 님의 말:

Jeroen Demeyer

unread,
Nov 23, 2013, 5:33:14 AM11/23/13
to sage-...@googlegroups.com
On 2013-11-23 06:52, R. Andrew Ohana wrote:
> They don't supply a Fortran compiler, which we need. Also as of Xcode 5
> they no longer supply gcc (the gcc in PATH is actually clang) and some
> packages we include depend explicitly on having gcc as our c compiler.
>
> If anything we need our own compiler now more than before, and this
> situation doesn't look like it will be changing given Apple's decisions.

See also http://trac.sagemath.org/ticket/12426
Reply all
Reply to author
Forward
0 new messages