Message-ID: <aW_Q6ixdtMqnzfhU@fedora>
Date: Tue, 20 Jan 2026 16:38:48 -0300
From: Wander Lairson Costa <wander@...hat.com>
To: Gabriele Monaco <gmonaco@...hat.com>
Cc: Steven Rostedt <rostedt@...dmis.org>, Nam Cao <namcao@...utronix.de>, 
	open list <linux-kernel@...r.kernel.org>, 
	"open list:RUNTIME VERIFICATION (RV)" <linux-trace-kernel@...r.kernel.org>
Subject: Re: [PATCH 17/26] rv/rvgen: fix possibly unbound variable in ltl2k

On Tue, Jan 20, 2026 at 01:30:35PM +0100, Gabriele Monaco wrote:
> On Tue, 2026-01-20 at 08:37 -0300, Wander Lairson Costa wrote:
> > On Tue, Jan 20, 2026 at 09:59:11AM +0100, Gabriele Monaco wrote:
> > > On Mon, 2026-01-19 at 17:45 -0300, Wander Lairson Costa wrote:
> > > > Initialize loop variable `i` before the for loop in abbreviate_atoms
> > > > function to fix pyright static type checker error. The previous code
> > > > left `i` potentially unbound in edge cases where the range could be
> > > > empty, though this would not occur in practice since the loop always
> > > > executes at least once with the given range parameters.
> > > > 
> > > > The initialization to zero ensures that `i` has a defined value before
> > > > entering the loop scope, satisfying static analysis requirements
> > > > while preserving the existing logic. The for loop immediately assigns
> > > > i to the first value from the range, so the initialization value is
> > > > never actually used in normal execution paths.
> > > > 
> > > > This change resolves the pyright reportPossiblyUnbound error without
> > > > altering the function's behavior or performance characteristics.
> > > 
> > > So are we just pleasing the tool or is there a real implication of this?
> > > 
> > > Apparently code like
> > > 
> > > for i in range(len([]), -1, -1):
> > >     pass
> > > print(i)
> > > 
> > > works just fine since range() returns at least 0 (as you mentioned in the
> > > commit
> > > message) and i is not used before assignation in the loop, so I don't really
> > > see
> > > a problem.
> > > 
> > > Apparently pyright devs don't want ([1]) to implement a logic to sort out
> > > the
> > > /possibly/ unbound error here.
> > > 
> > > From what I understand, this code is already not pythonic, so rather than
> > > silence the warning to please this tool I'd just refactor the code not to
> > > use i
> > > after the loop (or leave it as it is, since it works fine).
> > > 
> > > What do you think?
> > 
> > You're right, I could have done:
> > 
> > for atom in reversed(atoms): ...
> > 
> 
> I'm missing what you mean with this, the range is iterating over the string
> representation of atom (in reverse) not the array of atoms.
> 

Sorry, I misinterpreted you previous comment and picked the wrong piece
of code.

Yes, the basic goal was to make pyright happy.

> You basically want i to be the length of the longest prefix common to at least
> another atom.
> 
> You could assign i to some python trick doing the exact same thing the loop
> does, like:
> 
>     i = next((i for i in range(len(atom), -1, -1)
>         if sum(a.startswith(atom[:i]) for a in atoms) > 1))
> 
> next() is basically doing the break at the first occurrence from the generator,
> just now your i doesn't live (only) inside the loop.
> 
> So now you save 2 lines and get any C developer scratch their head when they
> look at the code, but hey, pyright is happy!
> 

Or just leave the assignment.

> If you do find the trick with next() readable or have any better idea, feel free
> to try though.
> 

Definitely the next() trick is not worth to make pyright happy.

> Thanks,
> Gabriele
> 
> > I will modify it in v2.
> > 
> > > 
> > > Thanks,
> > > Gabriele
> > > 
> > > [1] - https://github.com/microsoft/pyright/issues/844
> > > 
> > > > 
> > > > Signed-off-by: Wander Lairson Costa <wander@...hat.com>
> > > > ---
> > > >  tools/verification/rvgen/rvgen/ltl2k.py | 1 +
> > > >  1 file changed, 1 insertion(+)
> > > > 
> > > > diff --git a/tools/verification/rvgen/rvgen/ltl2k.py
> > > > b/tools/verification/rvgen/rvgen/ltl2k.py
> > > > index fa9ea6d597095..94dc64af1716d 100644
> > > > --- a/tools/verification/rvgen/rvgen/ltl2k.py
> > > > +++ b/tools/verification/rvgen/rvgen/ltl2k.py
> > > > @@ -45,6 +45,7 @@ def abbreviate_atoms(atoms: list[str]) -> list[str]:
> > > >  
> > > >      abbrs = []
> > > >      for atom in atoms:
> > > > +        i = 0
> > > >          for i in range(len(atom), -1, -1):
> > > >              if sum(a.startswith(atom[:i]) for a in atoms) > 1:
> > > >                  break
> > > 
> 

Powered by blists - more mailing lists