[sage-devel] Re: macports
Hi Sorry I've been unable to participate in the discussion, and I unfortunately won't get any time to look at for a while (we're moving city and jobs) so I won't say much because I won't be doing much for a couple of months. Really, my only concern is that sage compiles out of the box for newbies - the fink / macports problem just keeps popping up and is IMHO an obstacle to adoption and gives a very bad first impression. I would prefer a better solution than mine, but I tend to think that mine is better than none - I would sacrifice experts' comfort. Somebody may have pointed it out, but there is an issue with the sage build system that means you can work around it by simply running 'make' a second time. D ps the relevant ticket is (from memory) probably 4130 because my patch was broken on OSX 10.4 M: +61 423 535 397 W: +61 2 6125 8260 On 07/10/2008, at 6:28, Georg S. Weber [EMAIL PROTECTED] wrote: OK. I reopened the corresponding trac ticket #4125 (that went into Sage 3.1.2.rc4), and asked for that patch to be reverted. The underlying problem might have been addressed anyway with the patch in trac ticket #4127 (that went into Sage 3.1.3.alpha0), but Michael Abshoff and David Philp should please have another look at the issue. If I can be of any help (testing, more discussion), just let me know, I've got both a PPC Mac and an Intel Mac, the former has fink installed on it (which is not detected by the script in question, as it does not seem to have changed my path variable). Cheers, gsw --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
I think the best solution, if possible, is to re-adjust the paths to remove macports/fink to build. However, as a short term solution, rather than just killing the build, maybe it should be a prompt warning the user that it might be a problem, and asking if they want to continue anyways. It is apparent that for some people things go just as planned, and for those who continue and experience issues it will be obvious (to them and us) why. - Robert On Oct 7, 2008, at 6:52 AM, David Philp wrote: Hi Sorry I've been unable to participate in the discussion, and I unfortunately won't get any time to look at for a while (we're moving city and jobs) so I won't say much because I won't be doing much for a couple of months. Really, my only concern is that sage compiles out of the box for newbies - the fink / macports problem just keeps popping up and is IMHO an obstacle to adoption and gives a very bad first impression. I would prefer a better solution than mine, but I tend to think that mine is better than none - I would sacrifice experts' comfort. Somebody may have pointed it out, but there is an issue with the sage build system that means you can work around it by simply running 'make' a second time. D ps the relevant ticket is (from memory) probably 4130 because my patch was broken on OSX 10.4 M: +61 423 535 397 W: +61 2 6125 8260 On 07/10/2008, at 6:28, Georg S. Weber [EMAIL PROTECTED] wrote: OK. I reopened the corresponding trac ticket #4125 (that went into Sage 3.1.2.rc4), and asked for that patch to be reverted. The underlying problem might have been addressed anyway with the patch in trac ticket #4127 (that went into Sage 3.1.3.alpha0), but Michael Abshoff and David Philp should please have another look at the issue. If I can be of any help (testing, more discussion), just let me know, I've got both a PPC Mac and an Intel Mac, the former has fink installed on it (which is not detected by the script in question, as it does not seem to have changed my path variable). Cheers, gsw --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
Hi,
a solution which removes unwanted directories from the paths(s)
would use something like (untested! I just googled
http://www.faqs.org/docs/abs/HTML/string-manipulation.html):
path_to_clean = $PATH
dylibpath_to_clean = $DYLD_LIBRARY_PATH
${path_to_clean///opt/local*:/}
${dylibpath_to_clean///opt/local*:/}
${path_to_clean///sw*:/}
${dylibpath_to_clean///sw*:/}
PATH = path_to_clean export PATH
DYLD_LIBRARY_PATH = dylibpath_to_clean export DYLD_LIBRARY_PATH
But as I have been seating myself between the chairs already by
reopening #4125, I kindly ask to use the opportonity of SD 10 to first
sort out what exactly a potential enhanced fix is to achieve, before
more work is invested.
Cheers,
gsw
--~--~-~--~~~---~--~~
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://www.sagemath.org
-~--~~~~--~~--~--~---
[sage-devel] Re: macports
Hi, it is relatively easy to include workarounds in the script machinery, and for DYLD_LIBRARY_PATH this should already be the case. The difficult thing is to test, whether the problems users report every now and then do really disappear. It seems that none of the Sage developers has had these problems on a machine on his own (including myself). There only are these bug reports 1.Sage does not build 2. More or less dicussion 3. Hey, try to move macports out of the way 4. Now it builds, thank you! But I do not know how to produce out of the boxa kind of doctest for this, ensuring that a potential fix really is a working fix. Cheers, gsw --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
On Oct 5, 2008, at 11:38 , William Stein wrote: On Sun, Oct 5, 2008 at 6:52 AM, David Harvey [EMAIL PROTECTED] wrote: On Oct 5, 2008, at 5:32 AM, Justin C. Walker wrote: On Oct 5, 2008, at 02:11 , David Philp wrote: Would it be excessively uncivilised to *automatically* change the path during the build process? Well, the kind of people who will come across it will be capable of changing their own path---and maybe sensitive. And, it might require root permissions. Changing the PATH and related environment variables does not require any permissions. However, these variables can be modified each time a shell starts up, depending on how the startup works, so that may not suffice. It's kind of tricky, but David's suggestion makes sense to me. I've worked around this by changing the settings in my startup file, and that may be uncivilized. If it can be done without modifying the local startup files, that would be ideal. Any other thoughts? Two other thoughts. (1) this is the first time ever that I can recall sage not building on my machine by simply issuing make. I think it's a Bad Thing that this now happens. It will happen to me every time I build sage because I do not want to remove MacPorts permanently. (2) Here's how I got around it. I looked at my environment by typing env. Then I manually typed in export PATH=, including all the directories in my original PATH minus the ones involving /opt/local. Then I ran make. This seems to work. But Justin suggests above that this might not be enough; perhaps other shells being created during the build process are using a path from some startup script, which I haven't changed. So even though I am the kind of person who is capable of changing my path, but I actually have no idea what I'm doing, and quite possibly tricked sage into doing the wrong thing anyway. We could change things so that the shell script I wrote that calls spkg-install (I think it's maybe sage-spkg or something) cleans the path. Does MacPorts also modify *LD_LIBRARY_PATH*? Justin -- if we make sure no environment variables reference /opt and /sw, might that be enough to imply that nothing in macports (or fink) is going to get pulled in during the build? That should suffice. If neither the PATH nor library variables refer to those directories, I don't think system commands will find them. You should also keep in mind /usr/local; there is no system-provided content in that directory, so anything there can gum up the works (e.g., a library in /usr/local/lib might depend on something in the MacPorts or Fink libraries). However, there are some packages that seem to insist on looking in /usr/local (libintl, for one). I am for creating a workaround like above rather than forcing the user to do something funny every time. I'm for this because at least MacPorts (and Fink) are sufficiently standard on OS X that we can figure out how to workaround them. Amen. It's sufficiently tricky and arcane that our users who only want to use Sage, and not get into the sysadmin business, could be turned off completely by that requirement. Justin -- Justin C. Walker, Curmudgeon-At-Large Director Institute for the Enhancement of the Director's Income Weaseling out of things is what separates us from the animals. Well, except the weasel. - Homer J Simpson --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
OK. I reopened the corresponding trac ticket #4125 (that went into Sage 3.1.2.rc4), and asked for that patch to be reverted. The underlying problem might have been addressed anyway with the patch in trac ticket #4127 (that went into Sage 3.1.3.alpha0), but Michael Abshoff and David Philp should please have another look at the issue. If I can be of any help (testing, more discussion), just let me know, I've got both a PPC Mac and an Intel Mac, the former has fink installed on it (which is not detected by the script in question, as it does not seem to have changed my path variable). Cheers, gsw --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
On Oct 5, 2008, at 02:11 , David Philp wrote: Would it be excessively uncivilised to *automatically* change the path during the build process? Well, the kind of people who will come across it will be capable of changing their own path---and maybe sensitive. And, it might require root permissions. Changing the PATH and related environment variables does not require any permissions. However, these variables can be modified each time a shell starts up, depending on how the startup works, so that may not suffice. It's kind of tricky, but David's suggestion makes sense to me. I've worked around this by changing the settings in my startup file, and that may be uncivilized. If it can be done without modifying the local startup files, that would be ideal. Any other thoughts? Justin -- Justin C. Walker, Curmudgeon-At-Large Director Institute for the Enhancement of the Director's Income Weaseling out of things is what separates us from the animals. Well, except the weasel. - Homer J Simpson --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
Would it be excessively uncivilised to *automatically* change the path during the build process? Well, the kind of people who will come across it will be capable of changing their own path---and maybe sensitive. And, it might require root permissions. D --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
On Oct 5, 2008, at 5:32 AM, Justin C. Walker wrote: On Oct 5, 2008, at 02:11 , David Philp wrote: Would it be excessively uncivilised to *automatically* change the path during the build process? Well, the kind of people who will come across it will be capable of changing their own path---and maybe sensitive. And, it might require root permissions. Changing the PATH and related environment variables does not require any permissions. However, these variables can be modified each time a shell starts up, depending on how the startup works, so that may not suffice. It's kind of tricky, but David's suggestion makes sense to me. I've worked around this by changing the settings in my startup file, and that may be uncivilized. If it can be done without modifying the local startup files, that would be ideal. Any other thoughts? Two other thoughts. (1) this is the first time ever that I can recall sage not building on my machine by simply issuing make. I think it's a Bad Thing that this now happens. It will happen to me every time I build sage because I do not want to remove MacPorts permanently. (2) Here's how I got around it. I looked at my environment by typing env. Then I manually typed in export PATH=, including all the directories in my original PATH minus the ones involving /opt/local. Then I ran make. This seems to work. But Justin suggests above that this might not be enough; perhaps other shells being created during the build process are using a path from some startup script, which I haven't changed. So even though I am the kind of person who is capable of changing my path, but I actually have no idea what I'm doing, and quite possibly tricked sage into doing the wrong thing anyway. 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://www.sagemath.org -~--~~~~--~~--~--~---
[sage-devel] Re: macports
On Sun, Oct 5, 2008 at 6:52 AM, David Harvey [EMAIL PROTECTED] wrote: On Oct 5, 2008, at 5:32 AM, Justin C. Walker wrote: On Oct 5, 2008, at 02:11 , David Philp wrote: Would it be excessively uncivilised to *automatically* change the path during the build process? Well, the kind of people who will come across it will be capable of changing their own path---and maybe sensitive. And, it might require root permissions. Changing the PATH and related environment variables does not require any permissions. However, these variables can be modified each time a shell starts up, depending on how the startup works, so that may not suffice. It's kind of tricky, but David's suggestion makes sense to me. I've worked around this by changing the settings in my startup file, and that may be uncivilized. If it can be done without modifying the local startup files, that would be ideal. Any other thoughts? Two other thoughts. (1) this is the first time ever that I can recall sage not building on my machine by simply issuing make. I think it's a Bad Thing that this now happens. It will happen to me every time I build sage because I do not want to remove MacPorts permanently. (2) Here's how I got around it. I looked at my environment by typing env. Then I manually typed in export PATH=, including all the directories in my original PATH minus the ones involving /opt/local. Then I ran make. This seems to work. But Justin suggests above that this might not be enough; perhaps other shells being created during the build process are using a path from some startup script, which I haven't changed. So even though I am the kind of person who is capable of changing my path, but I actually have no idea what I'm doing, and quite possibly tricked sage into doing the wrong thing anyway. We could change things so that the shell script I wrote that calls spkg-install (I think it's maybe sage-spkg or something) cleans the path. Does MacPorts also modify *LD_LIBRARY_PATH*? Justin -- if we make sure no environment variables reference /opt and /sw, might that be enough to imply that nothing in macports (or fink) is going to get pulled in during the build? I am for creating a workaround like above rather than forcing the user to do something funny every time. I'm for this because at least MacPorts (and Fink) are sufficiently standard on OS X that we can figure out how to workaround them. William --~--~-~--~~~---~--~~ 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://www.sagemath.org -~--~~~~--~~--~--~---
