[sage-devel] Re: cannot build 2.8.4 on sage.math
Well, good news. I can tell you exactly what is causing the problem, and there are two easy ways to fix it. Here's what's going on: c_lib/ ntl_wrap.os is getting shipped with sage-2.8.4.tar, and that file was compiled with a compiler that includes the gnu stack-checking stuff (what they were talking about in the google article you found). Since that file exists, scons doesn't rebuild it when we do scons install, so it gets linked against into libcsage.so, which then gets put into $SAGE_ROOT/local/lib, and causes all your headache. You can even see this: do nm ntl_wrap.os | grep stack and you'll see the symbol there. So, your fix is easy: cd c_lib ; rm -f ntl_wrap.os ; scons install ... that will fix you up. There are two steps we're going to take to fix this. First, we'll make sure to remove the .os files in the sage-*.tar we distribute from now on, which will stop this from happening. Also, as a second check, I'm going to make sure that the sage install does something to remove those files on install, since there's no reason not to (if nothing else, simply to avoid annoying things like this). There's even a trac ticket, #626. I'm going to fix up the sage install process right now, and post a patch. I want to point out that this is probably the third or fourth weird build problem we've had with switching to scons, and even given all that, I still strongly support using scons, because I think it's worth it in the long run. I haven't used it too much (just for this, actually), but I've been very happy with it, and I think it's better for us than make. There are just some growing pains to be dealt with. ;) -cc On Sep 8, 2007, at 10:05 AM, David Harvey wrote: On Sep 8, 2007, at 1:03 PM, mabshoff wrote: Ok, I get loads of test failures on sage.math complaining about __stack_chk_fail. That is gmp related. I am looking trying to use the old p9 gmp package and see what will happen then. Great, so it's not just me then. I was really starting to go nuts. david --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
Hi David, I think this may be something in your ambient environment which scons handles differently than autoconf and friends. You could try replacing *** env = Environment() *** with env = Environment(ENV = os.environ) *** near the top of the c_lib/SConstruct file. SCons does not copy environment variables into it's environment. I think it is a strange design, but the SCons developers are quite dogmatic about it. This fixed a build problem for Kiran with thread title 2.8.3.x build problem. I'm not sure what that SConstruct looks like in 2.8.4, but I don't think this change made it in. I wasn't quite certain that I recommended the change, but if it fixes your problem as well, I'd say we should put it in. -- Joel On Saturday 08 September 2007 10:44, David Harvey wrote: I am tearing my hair out. I do a clean build of sage 2.8.4 on sage.math, and when I run it I get this: == -- | SAGE Version 2.8.4, Release Date: 2007-09-07 | | Type notebook() for the GUI, and license() for information.| -- --- type 'exceptions.ImportError' Traceback (most recent call last) /home/dmharvey/sage-2.8.4/local/bin/string in module() /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ preparser_ipython.py in module() 6 ### 7 8 import sage.misc.interpreter 9 10 import preparser /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ interpreter.py in module() 102 103 import os -- 104 import log 105 106 import remote_file /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ log.py in module() 51 52 import interpreter --- 53 import latex 54 import misc 55 /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ latex.py in module() 41 import random 42 --- 43 import sage.plot.all 44 45 from misc import tmp_dir /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ all.py in module() 2 text, circle, disk, hue, graphics_array, 3 list_plot, networkx_plot, parametric_plot, 4 polar_plot, contour_plot, arrow, 5 plot_vector_field, matrix_plot, bar_chart, 6 is_Graphics, /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ plot.py in module() 126 127 -- 128 from sage.structure.sage_object import SageObject 129 130 ## IMPORTANT: Do *not* import matplotlib at module scope. It takes a type 'exceptions.ImportError': /home/dmharvey/sage-2.8.4/local/lib/ libcsage.so: undefined symbol: __stack_chk_fail WARNING: Failure executing code: 'import sage.misc.preparser_ipython; sage.misc.preparser_ipython.magma_colon_equals=True' --- type 'exceptions.ImportError' Traceback (most recent call last) /home/dmharvey/sage-2.8.4/local/bin/ipython console in module() /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ interpreter.py in module() 102 103 import os -- 104 import log 105 106 import remote_file /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ log.py in module() 51 52 import interpreter --- 53 import latex 54 import misc 55 /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ latex.py in module() 41 import random 42 --- 43 import sage.plot.all 44 45 from misc import tmp_dir /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ all.py in module() 2 text, circle, disk, hue, graphics_array, 3 list_plot, networkx_plot, parametric_plot, 4 polar_plot, contour_plot, arrow, 5 plot_vector_field, matrix_plot, bar_chart, 6 is_Graphics, /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ plot.py in module() 126 127 -- 128 from sage.structure.sage_object import SageObject 129 130 ## IMPORTANT: Do *not* import matplotlib at module scope. It takes a type 'exceptions.ImportError': /home/dmharvey/sage-2.8.4/local/lib/ libcsage.so: undefined symbol: __stack_chk_fail
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 2007, at 11:00 AM, Joel B. Mohler wrote: Hi David, I think this may be something in your ambient environment which scons handles differently than autoconf and friends. Is ambient environment a technical unix term? Is that the same as my environment? You could try replacing *** env = Environment() *** with env = Environment(ENV = os.environ) *** near the top of the c_lib/SConstruct file. SCons does not copy environment variables into it's environment. I think it is a strange design, but the SCons developers are quite dogmatic about it. This fixed a build problem for Kiran with thread title 2.8.3.x build problem. I'm not sure what that SConstruct looks like in 2.8.4, but I don't think this change made it in. I wasn't quite certain that I recommended the change, but if it fixes your problem as well, I'd say we should put it in. It's already got that line in 2.8.4. Has anyone else actually built sage-2.8.4 from scratch on sage.math? david --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 5:09 pm, David Harvey [EMAIL PROTECTED] wrote: On Sep 8, 2007, at 11:00 AM, Joel B. Mohler wrote: Hello, Hi David, I think this may be something in your ambient environment which scons handles differently than autoconf and friends. Is ambient environment a technical unix term? Is that the same as my environment? ;), but I don't see anything suspicious. You could try replacing *** env = Environment() *** with env = Environment(ENV = os.environ) *** near the top of the c_lib/SConstruct file. SCons does not copy environment variables into it's environment. I think it is a strange design, but the SCons developers are quite dogmatic about it. This fixed a build problem for Kiran with thread title 2.8.3.x build problem. I'm not sure what that SConstruct looks like in 2.8.4, but I don't think this change made it in. I wasn't quite certain that I recommended the change, but if it fixes your problem as well, I'd say we should put it in. It's already got that line in 2.8.4. I would suggest commenting it out then. It was merged because of a build problem on RHEL 5. Has anyone else actually built sage-2.8.4 from scratch on sage.math? I believe William did do, I didn't try in person. david Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 2007, at 11:21 AM, mabshoff wrote: It's already got that line in 2.8.4. I would suggest commenting it out then. It was merged because of a build problem on RHEL 5. Okay, I'll try again with that line restored to how it was before. Has anyone else actually built sage-2.8.4 from scratch on sage.math? I believe William did do, I didn't try in person. Sorry I should rephrase that question: has anyone else actually built 2.8.4 and succesfully *run* it on sage.math? (The build itself appears to go smoothly.) david --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 5:23 pm, David Harvey [EMAIL PROTECTED] wrote: On Sep 8, 2007, at 11:21 AM, mabshoff wrote: It's already got that line in 2.8.4. I would suggest commenting it out then. It was merged because of a build problem on RHEL 5. Okay, I'll try again with that line restored to how it was before. Has anyone else actually built sage-2.8.4 from scratch on sage.math? I believe William did do, I didn't try in person. Sorry I should rephrase that question: has anyone else actually built 2.8.4 and succesfully *run* it on sage.math? (The build itself appears to go smoothly.) I am sure William ran testall, but I just started a build to see if I can replicate your problem. I shall report back in about 2 hours (unless something big is running on sage.math). david Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Saturday 08 September 2007 11:09, David Harvey wrote: I think this may be something in your ambient environment which scons handles differently than autoconf and friends. Is ambient environment a technical unix term? Is that the same as my environment? No, sorry. I was intending it as a bit more generic than the technical term environment. The bit that I understood on the similar IRC conversation seemed like you were linking to a gmp outside your sage tree. Now, how that would have occurred is probably a matter of the technical-term environment. -- Joel --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
Yeah, that doesn't sound like fun. Try the following from your sage- main directory: cd c_lib ; rm -f libcsage.* ; scons install. You should see scons rebuild the libcsage.so and then copy it to your $SAGE_ROOT/local/lib ... let me know if that works. If it does, we might just need to be more adamant about how the c_lib build happens on install, because it sounds like your c_lib isn't getting rebuilt (i.e. your $SAGE_ROOT/local/lib/libcsage.so is bunk). In fact, I can see this being pretty likely -- it could be the same problem I was having when I switched branches, and I just need to make sure the same thing is happening when you do an install. This shouldn't be too bad, and it'll save you from having to pull out any more hair. ;) If *not*, it means there's actually a problem with the libcsage.so that's getting built, which is bad. This will require more thought. I'll be gone most of the day, but I'll read your reply when I get back and make a fix if it's the first one. -cc On Sep 8, 2007, at 7:44 AM, David Harvey wrote: I am tearing my hair out. I do a clean build of sage 2.8.4 on sage.math, and when I run it I get this: == -- | SAGE Version 2.8.4, Release Date: 2007-09-07 | | Type notebook() for the GUI, and license() for information.| -- -- -- --- type 'exceptions.ImportError' Traceback (most recent call last) /home/dmharvey/sage-2.8.4/local/bin/string in module() /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ preparser_ipython.py in module() 6 ## ## ### 7 8 import sage.misc.interpreter 9 10 import preparser /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ interpreter.py in module() 102 103 import os -- 104 import log 105 106 import remote_file /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ log.py in module() 51 52 import interpreter --- 53 import latex 54 import misc 55 /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ latex.py in module() 41 import random 42 --- 43 import sage.plot.all 44 45 from misc import tmp_dir /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ all.py in module() 2 text, circle, disk, hue, graphics_array, 3 list_plot, networkx_plot, parametric_plot, 4 polar_plot, contour_plot, arrow, 5 plot_vector_field, matrix_plot, bar_chart, 6 is_Graphics, /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ plot.py in module() 126 ## ## 127 -- 128 from sage.structure.sage_object import SageObject 129 130 ## IMPORTANT: Do *not* import matplotlib at module scope. It takes a type 'exceptions.ImportError': /home/dmharvey/sage-2.8.4/local/lib/ libcsage.so: undefined symbol: __stack_chk_fail WARNING: Failure executing code: 'import sage.misc.preparser_ipython; sage.misc.preparser_ipython.magma_colon_equals=True' -- -- --- type 'exceptions.ImportError' Traceback (most recent call last) /home/dmharvey/sage-2.8.4/local/bin/ipython console in module() /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ interpreter.py in module() 102 103 import os -- 104 import log 105 106 import remote_file /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ log.py in module() 51 52 import interpreter --- 53 import latex 54 import misc 55 /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/misc/ latex.py in module() 41 import random 42 --- 43 import sage.plot.all 44 45 from misc import tmp_dir /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ all.py in module() 2 text, circle, disk, hue, graphics_array, 3 list_plot, networkx_plot, parametric_plot, 4 polar_plot, contour_plot, arrow, 5 plot_vector_field, matrix_plot, bar_chart, 6 is_Graphics, /home/dmharvey/sage-2.8.4/local/lib/python2.5/site-packages/sage/plot/ plot.py in module() 126 ## ##
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 6:38 pm, Craig Citro [EMAIL PROTECTED] wrote: Yeah, that doesn't sound like fun. Try the following from your sage- main directory: cd c_lib ; rm -f libcsage.* ; scons install. You should see scons rebuild the libcsage.so and then copy it to your $SAGE_ROOT/local/lib ... let me know if that works. If it does, we might just need to be more adamant about how the c_lib build happens on install, because it sounds like your c_lib isn't getting rebuilt (i.e. your $SAGE_ROOT/local/lib/libcsage.so is bunk). In fact, I can see this being pretty likely -- it could be the same problem I was having when I switched branches, and I just need to make sure the same thing is happening when you do an install. This shouldn't be too bad, and it'll save you from having to pull out any more hair. ;) If *not*, it means there's actually a problem with the libcsage.so that's getting built, which is bad. This will require more thought. I'll be gone most of the day, but I'll read your reply when I get back and make a fix if it's the first one. -cc On Sep 8, 2007, at 7:44 AM, David Harvey wrote: SNIP Ok, I get loads of test failures on sage.math complaining about __stack_chk_fail. That is gmp related. I am looking trying to use the old p9 gmp package and see what will happen then. Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 2007, at 1:03 PM, mabshoff wrote: Ok, I get loads of test failures on sage.math complaining about __stack_chk_fail. That is gmp related. I am looking trying to use the old p9 gmp package and see what will happen then. Great, so it's not just me then. I was really starting to go nuts. david --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 8, 7:05 pm, David Harvey [EMAIL PROTECTED] wrote: On Sep 8, 2007, at 1:03 PM, mabshoff wrote: Hey David, Ok, I get loads of test failures on sage.math complaining about __stack_chk_fail. That is gmp related. I am looking trying to use the old p9 gmp package and see what will happen then. Great, so it's not just me then. I was really starting to go nuts. :) - I did install gmp-4.2.1p9, did a sage -ba a now I no longer see the problem about __stack_chk_fail. Some doctest fail, but that is to be expected because the we get different but correct output for various gmp related functions like crt. Now the interesting question: Why does the new spkg fail on sage.math? We did test it on many, many systems. Oh well. I will run make check on both spkgs and see if anything pops up. david Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
Why does Sage 2.8.4 crash on sage.math? Executive Summary: SCons somehow gets its LD_LIBRARY_PATH or env in general wrong and ends up linking against the gmp 4.2.1 provided by Debian. Somehow the compiler gets it all wrong and *boom* How to fix: Fix SCons *ducks* - a short term solution seems to be to source local/sage/env before building sage-2.8.4.spkg. The detailed story: untar sage 2.8.4, make, wait a while. Then: [EMAIL PROTECTED]:/tmp/Work-mabshoff/sage-2.8.4-2$ ldd local/lib/ libcsage.so libntl.so = not found libgmp.so.3 = /usr/lib/libgmp.so.3 (0x2abf99d8d000) libstdc++.so.6 = /usr/lib/libstdc++.so.6 (0x2abf99ecb000) libm.so.6 = /lib/libm.so.6 (0x2abf9a0c9000) libgcc_s.so.1 = /lib/libgcc_s.so.1 (0x2abf9a24c000) libc.so.6 = /lib/libc.so.6 (0x2abf9a359000) /lib64/ld-linux-x86-64.so.2 (0x4000) That doesn't look good. start Sage: *boom* [EMAIL PROTECTED]:/tmp/Work-mabshoff/sage-2.8.4-2$ ./sage -- | SAGE Version 2.8.4, Release Date: 2007-09-07 | | Type notebook() for the GUI, and license() for information.| -- --- type 'exceptions.ImportError' Traceback (most recent call last) /tmp/Work-mabshoff/sage-2.8.4-2/local/bin/string in module() /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/python2.5/site-packages/sage/ misc/preparser_ipython.py in module() 6 ### 7 8 import sage.misc.interpreter 9 10 import preparser SNIP /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/python2.5/site-packages/sage/ plot/all.py in module() 2 text, circle, disk, hue, graphics_array, 3 list_plot, networkx_plot, parametric_plot, 4 polar_plot, contour_plot, arrow, 5 plot_vector_field, matrix_plot, bar_chart, 6 is_Graphics, /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/python2.5/site-packages/sage/ plot/plot.py in module() 126 127 -- 128 from sage.structure.sage_object import SageObject 129 130 ## IMPORTANT: Do *not* import matplotlib at module scope. It takes a type 'exceptions.ImportError': /tmp/Work-mabshoff/sage-2.8.4-2/local/ lib/libcsage.so: undefined symbol: __stack_chk_fail ERROR: name 'sage_prompt' is not defined source local/bin/sage-env - Now it looks a little better: [EMAIL PROTECTED]:/tmp/Work-mabshoff/sage-2.8.4-2$ source local/bin/sage- env [EMAIL PROTECTED]:/tmp/Work-mabshoff/sage-2.8.4-2$ ldd local/lib/ libcsage.so libntl.so = /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/ libntl.so (0x2b67db845000) libgmp.so.3 = /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/ libgmp.so.3 (0x2b67dbb0c000) libstdc++.so.6 = /usr/lib/libstdc++.so.6 (0x2b67dbc5f000) libm.so.6 = /lib/libm.so.6 (0x2b67dbe5d000) libgcc_s.so.1 = /lib/libgcc_s.so.1 (0x2b67dbfe) libc.so.6 = /lib/libc.so.6 (0x2b67dc0ed000) /lib64/ld-linux-x86-64.so.2 (0x4000) But still: SNIP type 'exceptions.ImportError': /tmp/Work-mabshoff/sage-2.8.4-2/local/ lib/libcsage.so: undefined symbol: __stack_chk_fail Now [having sourced local/bin/sage-env]: ./sage -ba [drink a couple of gin tonics/cups of coffee/tea ... whatever] SNIP running install_scripts changing mode of /tmp/Work-mabshoff/sage-2.8.4-2/local/bin/ dsage_server.py to 755 changing mode of /tmp/Work-mabshoff/sage-2.8.4-2/local/bin/ dsage_worker.py to 755 changing mode of /tmp/Work-mabshoff/sage-2.8.4-2/local/bin/ dsage_setup.py to 755 running install_egg_info Removing /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/python2.5/site- packages/sage-0.0.0-py2.5.egg-info Writing /tmp/Work-mabshoff/sage-2.8.4-2/local/lib/python2.5/site- packages/sage-0.0.0-py2.5.egg-info -- | SAGE Version 2.8.4, Release Date: 2007-09-07 | | Type notebook() for the GUI, and license() for information.| -- sage: Exiting SAGE (CPU time 0m0.00s, Wall time 0m4.86s). If anybody wants to make a proper patch, open a ticket and send a patch to William please do so. I gotta get back to fixing Conditional jump or move depends on uninitialised value(s) in LinBox :) Thank you, come again. Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Saturday 08 September 2007 17:49, mabshoff wrote: Executive Summary: SCons somehow gets its LD_LIBRARY_PATH or env in general wrong and ends up linking against the gmp 4.2.1 provided by Debian. Somehow the compiler gets it all wrong and *boom* How to fix: Fix SCons *ducks* - a short term solution seems to be to source local/sage/env before building sage-2.8.4.spkg. This confuses me. How is source'ing local/sage/env any different from what normally happens? Isn't that always source'd once you get into the internals of sage batch files? So, I thought I could answer that question myself: In local/bin/sage-build we call local/bin/sage-env with-out source, but inside of local/bin/sage-env, we say you must: # You must source this; see below! Thus, I'm thinking that line 45 of sage-build needs 'source ' at the front, but I'm too close to my bed-time to actually go test that myself. I believe that autoconf might have not cared so much about things like this because the --prefix option got set which essentially rewrote the environment on it's own. -- Joel --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: cannot build 2.8.4 on sage.math
On Sep 9, 3:10 am, mabshoff [EMAIL PROTECTED] dortmund.de wrote: On Sep 9, 2:40 am, Joel B. Mohler [EMAIL PROTECTED] wrote: On Saturday 08 September 2007 17:49, mabshoff wrote: Executive Summary: SCons somehow gets its LD_LIBRARY_PATH or env in general wrong and ends up linking against the gmp 4.2.1 provided by Debian. Somehow the compiler gets it all wrong and *boom* How to fix: Fix SCons *ducks* - a short term solution seems to be to source local/sage/env before building sage-2.8.4.spkg. Hello Joel, This confuses me. Oh, me too. How is source'ing local/sage/env any different from what normally happens? Are you saying that in a source blah vs ./blah way or more meta/ abstract? Isn't that always source'd once you get into the internals of sage batch files? I don't know. So, I thought I could answer that question myself: In local/bin/sage-build we call local/bin/sage-env with-out source, but inside of local/bin/sage-env, we say you must: # You must source this; see below! Thus, I'm thinking that line 45 of sage-build needs 'source ' at the front, but I'm too close to my bed-time to actually go test that myself. I did that, started a build and will see in the morning - or in about 90 minutes depending if I stay motivated that long and will let you know. Sorry, but that didn't work. I am out of ideas for the night. Let's hope somebody else can pick it up from here. Cheers, Michael I believe that autoconf might have not cared so much about things like this because the --prefix option got set which essentially rewrote the environment on it's own. Well, SCons is without a doubt better than autohell, we just have to get through the teething phase until we are good to go. -- Joel Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---