Hey Nathann,
On Thursday, December 17, 2015 at 8:33:48 PM UTC-6, Nathann Cohen wrote:
>
> > That would completely over-engineer things, be backwards incompatible,
> and
> > would make users hate us. This is why we have exceptions.
>
> Stick to that "would make users hate us" thought and look at what we have:
>
> - A method that "sometimes work, and sometimes does not", because it
> does not apply to all instances of its class
>
It is perfectly valid for a method to raise an error if the current
instance does not meet certain criteria. I feel a double standard here with
multiedges and loops...
>
> - Mathematically speaking, this method cannot even be *defined* on the
> object to which it belongs (the order does not always exist)
>
It has a perfectly valid and useful definition with a very mild assumption.
>
> - No documentation of whenever this function can be expected to work
> (in particular no warning)
>
I don't mind adding a *little* warning to the doc.
>
> - Unclear error message when it fails:
>
> sage: SetPartition([{'a','b','c'}]).to_permutation()
> ...
> TypeError: cannot concatenate 'str' and 'int' objects
>
This will be fixed by the warning above. Although I would actually argue
that this implementation cuts a few too many corners in this case.
>
> If what I propose is "over engineering", do you have anything better?
>
> One or two warnings along with a little bit of coding to improve things
without throwing fud all over something.
BTW, the only messages complaining about this I've seen is from you.
Best,
Travis
--
You received this message because you are subscribed to the Google Groups
"sage-devel" 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/sage-devel.
For more options, visit https://groups.google.com/d/optout.
