I noticed that polyvar-ids are generated using gensym.
For me, this is not very convenient for two reasons:
1.) I am using the atom-editor and every time I reload the shell, a new instance of variables is created, which can then cause unexpected errors when suddenly two polyvars with the same name are not considered equal.
2.) I am distributing my algorithm among different cores and when I do something like @everywhere @ncpolyvar x, the workers might not have the same object of x which makes it more difficult to share tasks.
What if we generate the ids uniquely by their name (without any "randomness" that gensym adds)? For which usecases would this cause problems?
I noticed that polyvar-ids are generated using
gensym
.For me, this is not very convenient for two reasons: 1.) I am using the atom-editor and every time I reload the shell, a new instance of variables is created, which can then cause unexpected errors when suddenly two polyvars with the same name are not considered equal. 2.) I am distributing my algorithm among different cores and when I do something like
@everywhere @ncpolyvar x
, the workers might not have the same object of x which makes it more difficult to share tasks.What if we generate the ids uniquely by their name (without any "randomness" that
gensym
adds)? For which usecases would this cause problems?