2011/3/4 Simon King <simon.k...@uni-jena.de>:
> PS:
>
> On 4 Mrz., 09:06, Simon King <simon.k...@uni-jena.de> wrote:
>> What about the following idea:
>> 1. var(s) only accept strings and lists of strings, and for backwards
>> compatibility one could also make it accept objects whose string
>> representation does not contain commas or whitespace. So,
>> var(QQ['t'].gen()) is fine, but with input like var(QQ['t']), an error
>> is raised.
>> 2. If the input is a valid identifier (all ascii) and not a reserved
>> Python keyword, it will be inserted into the global namespace.
>> 3. Otherwise, a variable with the given name is returned, but *not*
>> put into the global namespace. We may think of printing a warning
>> message in that case. In that way, alpha=var("α") would still work.
>
> No, that's a bad idea. It wouldn't change the current situation at
> all. In particular, sin([])=[] would still "work" (but shouldn't).
>
> Perhaps the following is better:
> Introduce an optional argument validate_name. If it is True (which is
> the default), an error is raised if the name is no valid identifier or
> if it is a reserved keyword. So, sin([])=[] or var('def') would result
> in an error.
> But if validate_name=False is used, then simply a variable of the
> given name is returned (and not inserted into the global name space).
So one would have to do
alpha=var("α", validate_name=False)
to avoid an error? Also, I'd still like to avoid
var("[]", ...)
working at all. We could accept all valid Python3 identifiers, and
print a warning if we were unable to insert them into the global
namespace, e.g. they're not valid Python2 identifiers. (The warning
could be suppressed if one did, say, var("α", inject=False).)
- Robert
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
