Is explicit reset of precision helpful?

[JeffreySarnoff]

Currently, the definitions use the precision set for BigFloat / MPFR as it happens to be when the package is usinged.
Does it make sense to include e.g. setprecision(BigFloat, max(precision(BigFloat), 256)) before the includes?

[devmotion]

I'm not sure I understand. IrrationalConstants should use exactly the same behaviour as Base when defining irrationals (just a separate type system to avoid type piracy which IMO should be adopted by Base), so in case there are issues with this behaviour I think they should be fixed upstream first.

[JeffreySarnoff]

That is (half of) the issue. Base uses setprecision(BigFloat, 256).
If I do this:

(somewhere else) setprecision(BigFloat, 128)
using IrrationalConstants

my irrational constants will be created with 128 BigFloats ..
so BigFloat(<irrationalconstant>) will have less precision than one gets with Base.

[devmotion]

The irrational constants are not defined as BigFloats, so I don't know what exactly you refer to when you write "my irrational constants will be created with 128 BigFloats". Irrational constants can be converted to BigFloats explicitly, in which case the precision by default will be set to precision(BigFloat) but can be specified explicitly as keyword argument precision.

[JeffreySarnoff]

ok --