Paul Rubin <http://[EMAIL PROTECTED]> wrote:
> Dave Benjamin <[EMAIL PROTECTED]> writes: > > Can we get a show of hands for all of those who have written or are > > currently maintaining code that uses the leaky listcomp "feature"? > > It's really irrelevant whether anyone is using a feature or not. If > the feature is documented as being available, it means that removing > it is an incompatible change that can break existing code which > currently conforms to the spec. If the "feature" is described as the > bug that it is, anything that relies on it is nonconformant. > > I do remember seeing some cute tricks (i.e. capable of becoming > idioms) that depend on the leakage. Sure, ``if [mo for mo in [myre.search(line)] if mo]: use(mo)`` and the like, used to simulate assign-and-test. No doubt maintaining backwards compat with this kind of horrors means listcomps need to keep leaking until Python 3.0, alas. But hopefully no farther... Alex -- http://mail.python.org/mailman/listinfo/python-list