#10903: Update Singular to 3-1-3-3.
------------------------+---------------------------------------------------
Reporter: malb | Owner: tbd
Type: defect | Status: needs_info
Priority: critical | Milestone: sage-4.7.2
Component: packages | Keywords: singular SageDays34 sd34
Work_issues: | Upstream: N/A
Reviewer: | Author: Burcin Erocal, Martin Albrecht,
Volker Braun
Merged: | Dependencies: #11339
------------------------+---------------------------------------------------
Comment(by vbraun):
The rules for using/setting `currRing` when calling libSingular are:
* The Sage library Python code should never make any assumptions about
`currRing`.
* `currRing` can only be assumed to stay defined within pure C/Cython
code.
This is particularly important because the garbage collector can run
between any two Python instructions, but not within C/Cython!
In order to identify code where we forget to set `currRing`, I implemented
a function `poison_currRing()` in `sage.libs.singular.singular` that will
set `currRing` to NULL, triggering a segfault if we forget to set it back
to a meaningful value before calling libSingular. Using the Python
debugger hook, this function can be called between each Python command
(but not within Cython code).
It can be globally enabled by uncommenting the lines
{{{
### Debugging for Singular, see trac #10903
from sage.libs.singular.ring import poison_currRing
sys.settrace(poison_currRing)
}}}
at the very end of `sage/all.py`.
Using this code, I identified and fixed various places in the Sage library
of the form
{{{
#!python
rChangeCurrRing(_ring)
self.parent().some_python_method()
libSingular_func()
}}}
Note that calling some python method in-between exits Cython and allows
Python to run the garbage collector etc! Not only do you exit Cython when
you return from your Cython function, but also if you call Python in-
between.
I also ran into an issue where code uses iteritems() to iterate over
`locals()`, `globals()`, or dictionaries returned by `gc.get_referrers()`.
This can potentially raise a `RuntimeError` if those dictionaries change
during the iteration. Hooking into the python debugger loop is a good way
of triggering precisely that. This also happens in the `PolyBoRi` Python
interface, this is now fixed upstream at
https://sourceforge.net/tracker/?func=detail&aid=3416946&group_id=210499&atid=1013986
With the attached patch, I can run the whole Sage testsuite without errors
while continuosly poisoning `currRing`. If you do not patch `PolyBoRi`,
you get some otherwise harmless errors from it but no segfaults. Note that
the patch does not enable the poisoning by default. I am quite sure that
the patch fixes all remaining Sage/Singular segfault issues. Please give
it a try!
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/10903#comment:42>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
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-trac?hl=en.
