That will be fine until FriCAS allows inherent parellelism. I don't think we have too many places where parellel execution of commands will fail, but marking domain variables that are not "constant" like here https://github.com/fricas/fricas/blob/master/src/algebra/expr.spad#L1084 seems to me like a good idea.
In fact, is it really clear that FriCAS accesses tmp0 from outside the function? If I'm not wrong, in Aldor you would have to declare free tmp0; inside subtractIfCan. Ralf On 03/26/2018 02:05 PM, oldk1331 wrote: > This change has been discussed long time ago, > now I'm posting it again for review: > > By using domain variables 'tmp0' and 'tmp1', > this patch avoids allocation when computation > fails. > > diff --git a/src/algebra/vector.spad b/src/algebra/vector.spad > index 6b6d7907..cb9e5b32 100644 > --- a/src/algebra/vector.spad > +++ b/src/algebra/vector.spad > @@ -437,24 +437,28 @@ > u : % * v : % == map(_*, u, v)$Rep > > if R has CancellationAbelianMonoid then > + tmp0 := new(dim, 0)$Rep > + > subtractIfCan(u:%, v:%):Union(%,"failed") == > - w := new(dim, 0)$Vector(R) > + w := tmp0 > for i in 1..dim repeat > (c := subtractIfCan(qelt(u, i)$Rep, qelt(v,i)$Rep)) case > "failed" => > return "failed" > qsetelt!(w, i, c::R)$Rep > - w pretend % > + copy w > > if R has Monoid then > > 1 == new(dim, 1)$Vector(R) pretend % > > + tmp1 := new(dim, 1)$Rep > + > recip z == > - w := new(dim, 1)$Vector(R) > - for i in minIndex w .. maxIndex w repeat > + w := tmp1 > + for i in 1..dim repeat > (u := recip qelt(z, i)) case "failed" => return "failed" > qsetelt!(w, i, u::R) > - w pretend % > + copy w > > if R has AbelianMonoid and R has Monoid then > -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To post to this group, send email to fricas-devel@googlegroups.com. Visit this group at https://groups.google.com/group/fricas-devel. For more options, visit https://groups.google.com/d/optout.
