Hmm, I see...so in theory, if I am writing a simple string parser, and out of convention have used malloc() before stumbling onto symem_newptr(), it shouldn't wreak any havoc in my routine. I was worried I may have missed something major.
The biggest argument I see for using sysmem_newptr / sysmem_freeptr is to prevent bugs. For example, it would be bad to call free() on an argv returned by object_attr_getvalueof(). So, it is not a bad habit to use the sysmem_* memory routines when programming for a max environment.
It is also possible that sysmem_newptr() could be faster than malloc() for many repeated small memory allocations.