A list for the developers of CellML tools
Text archives Help
[cellml-dev] It seems that after folding together adjacent units of the same base type with exponents that do not sum to zero, we were referencing a freed pointer when canonicalising units; uLast should be add_ref'ed when we create a new CDABaseUnitInstance.
Chronological Thread
- From: ak.miller at auckland.ac.nz (Andrew Miller)
- Subject: [cellml-dev] It seems that after folding together adjacent units of the same base type with exponents that do not sum to zero, we were referencing a freed pointer when canonicalising units; uLast should be add_ref'ed when we create a new CDABaseUnitInstance.
- Date: Tue, 01 Dec 2009 19:45:08 +1300
Hi Justin,
Is there a specific example of a model where this is an issue? The
reference count starts at one when an object is created, so in most
cases this is going to create a memory leak. I'm a bit concerned that
this could simply be 'upping the reference count by one' to fix a
problem somewhere else - which is really just creating another bug to
cancel out the first one under some circumstances. I suggest that if
there is a model which causes problem under the code, a tracker item be
created for it.
><http://cellml-api.hg.sourceforge.net/hgweb/cellml-api/cellml-api/rev/42412b41ed02>
Best wishes,
Andrew
- [cellml-dev] It seems that after folding together adjacent units of the same base type with exponents that do not sum to zero, we were referencing a freed pointer when canonicalising units; uLast should be add_ref'ed when we create a new CDABaseUnitInstance., Andrew Miller, 12/01/2009
Archive powered by MHonArc 2.6.18.
