sys.exc_clear() does not seem to exist in Python 3.0 anymore, so, a way
to deal with it should be provided (maybe put a #TODO comment and point
to somewhere explaining what happened?).

There should be no need for sys.exc_clear() anymore. What are you trying
to achieve?

In any case, probably a documentation issue :))

exc_clear() is not in the 3.0 docs, and the whatsnew contains a note.

Maybe a 2to3 issue?

When created it was already marked as a 2to3 issue (in the components),
so, for clarity, yes: it's a 2to3 issue (what should the user do with
that when porting... I think that the 2to3 should do something regarding
that... maybe just changing it for a 'pass'?).

Also, the docs say it was removed, but they don't leave clear that it's
not needed because no info is stored anymore (because that was needed
just to clear the frame that was kept alive when an exception was raised
-- so, note that I'm assuming that issue was fixed and it's not needed
anymore, but the docs don't make that clear)

Le lundi 22 décembre 2008 à 12:48 +0000, Fabio Zadrozny a écrit :
> Fabio Zadrozny <fabioz@users.sourceforge.net> added the comment:
> 
> When created it was already marked as a 2to3 issue

Oops, sorry.

> (because that was needed
> just to clear the frame that was kept alive when an exception was raised

To clear the frame and the exception object itself, yes.

Actually, I think this should just get a Py3k warning in 2.6 and 2.7.
2to3 doesn't deal with anything else that has been removed.

> 2to3 doesn't deal with anything else that has been removed.

That seems a bit odd for me... isn't it the perfect place for that? (it
doesn't even need to change the code for a pass, but it could give the
user some warning about it at that specific line -- maybe with a pointer
to a place explaining why it was removed).

I think that if 2to3 flags it, it can make the porting cycle faster than
having to run the code (and expecting a huge code-coverage) to get those
-- especially as sys.exc_clear is mostly used in cases where you are
expecting some exception... and I believe not everyone gives as much
emphasis for exception cases as they'd for the cases in the 'regular flow'

In fact, there is already a Py3k warning for sys.exc_clear(). I'm still
not convinced that 2to3 should play a role. What should it do? Simply
remove the call? I'm apprehensive about trying to guess the intention of
the user.

If this is already handled by the -3 flag then I agree it's not a bug.