On Fri, Feb 20, 2009 at 8:34 PM, mabshoff <[email protected]> wrote:
>
>
>
> On Feb 20, 8:14 pm, William Stein <[email protected]> wrote:
>> On Fri, Feb 20, 2009 at 4:58 PM, William Stein <[email protected]> wrote:
>> > Hi,
>>
>> > I did have at least two boxes where Atlas fails to build.  Is there a
>> > flag I have to set so it auto-restarts??
>>
>> > chmod: cannot access
>> > `/space/wstein/farm/sage-3.3.rc3/local/lib/libptf77blas.a': No such
>> > file or directory
>
> This error is completely normal since we do not build the multi
> threaded libraries.

OK, this doesn't matter.  But the build does fail.

>
>> > make[3]: [install_lib] Error 1 (ignored)
>> > make[3]: Leaving directory
>> > `/space/wstein/farm/sage-3.3.rc3/spkg/build/atlas-3.8.3/ATLAS-build'
>> > make[2]: Leaving directory
>> > `/space/wstein/farm/sage-3.3.rc3/spkg/build/atlas-3.8.3/ATLAS-build'
>> > ATLAS failed to build because your system is too heavily loaded to
>> > obtain accurate timing.
>> > Please restart the build by typing make, when the load on your system
>> > has decreased.
>>
>> > real    19m32.137s
>> > user    11m29.219s
>> > sys     2m58.263s
>> > sage: An error occurred while installing atlas-3.8.3
>> > Please email sage-develhttp://groups.google.com/group/sage-devel
>>
>> I've re-opened trachttp://trac.sagemath.org/sage_trac/ticket/1641
>> which is about auto-restarting the build, and posted some comments
>> on that ticket.
>
> Ok. The suggestion you made about restarting spkg-install being called
> repeatedly does not work since you either need to call configure with
> a special parameter or call make again.

How can running the current spkg-install from scratch again possibly
not work?  Well I implemented it (see spkg note below), and am testing
it now to see if it works.

> The restart only happens up to
> five times - anything else and your machine is misconfigured.

In the above example the restart happened exactly once, then the build
failed later and I got an error about the machine being too loaded.
The script thus didn't successfully do the job it is supposed to do.

>> Other than this, Sage is in very good shape.  Except for the boxes
>> where sage failed to build because of Atlas, all standard doctests
>> pass.  That #long OS X doctest in ell_finite_field.py still fails, of
>> course, but that's a basic limitation until libsingular is improved.
>> And even #1641 being a problem isn't a show stopper, since it's an
>> enhancement.
>>
>> I refereed #5318 which should definitely go in.  It is a major
>> performance mistake in the new multiedges digraph plotting code that
>> came up during the status reports today.
>
> Ok.
>
>> I'm working on redoing #1641 -- the ATLAS autorestart -- using a more
>> sophisticated script right now.  I'll post a patch if I can get it to
>> work and pass my rigorous test (building 12 ATLAS's at once on
>> boxen.math).
>
> Where are the build failure logs?

/home/wstein/farm/out/3.3.rc3/

The new spkg I'm working on is:

  /home/wstein/farm/atlas-3.8.3.p1.spkg

I don't claim it works since the "rigorous testing" I'm done hasn't
finished... and you know what they say -- if it isn't tested then it
is broken.

William

--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to [email protected]
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
-~----------~----~----~----~------~----~------~--~---

Reply via email to