On 09/18/2017 01:19 AM, oldk1331 wrote: > I think the compiler treat SubDomain and its argument specially, > it only handles INT/NNI/PI, so I suggest you to not use SubDomain.
I don't know how the compiler treats SubDomain, but I find it logically questionable. Think about a domain T that is an OrderedAbelianGroup. Then what do you expect the Domain D := SubDomain(T, 0< #1) to export. Does it export -: (%, %) -> % like T? Or is it (%, %) -> Union(%, "failed") in D? Why would one expect a change of the signature? And even if it is -: (%, %) -> % in D, then it is most probably only a partial function. So the ++ docstrings would document something that does not fit the implementation. Would you expect D to export the function -: % -> % at all? If there were no support from the compiler for NNI and PI, I don't think that for user-defined domains such questions lead to a reasonable compiled code. So I also support Qian. We should rather get rid of SubDomain than proposing to use it. But since it is only used for PI and NNI and makes no (big) problems there, removal is not firs priority. Ralf -- 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 [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/fricas-devel. For more options, visit https://groups.google.com/d/optout.
