Domain errors based on type do not make sense for the core language (primitives as opposed to foreigns - type is a foreign issue).
Domain errors should be based on domain, not on type. If we wanted types to matter in the core language, we should have type errors. (But we should not want that, either.) Thanks, -- Raul On Fri, Aug 26, 2016 at 4:54 PM, robert therriault <[email protected]> wrote: > Looks like u: has a different domain than 10&u: which is interesting as I am > not sure what meaning a floating or complex argument would have for u: > > Makes more sense to me to have it return a domain error based on type than > the present behaviour, but backwards compatibility could always be a factor. > > JVERSION > Engine: j805/j64/darwin > Beta-11: commercial/2016-08-23T13:05:27 > Library: 8.05.03 > Qt IDE: 1.4.10/5.4.2 > Platform: Darwin 64 > Installer: J805 install > InstallPath: /applications/j64-80511 > Contact: www.jsoftware.com > u: {.50 1.1 > 2 > u: {. 50 1j1 > 2 > 10 u: {.50 1.1 > |domain error > | 10 u:{.50 1.1 > u: 1j1 > |domain error > | u:1j1 > > Cheers, bob > >> On Aug 26, 2016, at 1:43 PM, Raul Miller <[email protected]> wrote: >> >> u: {.50 1.1 > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
