Regression: many2one search operator 'in'

Hello JFB,

Thanks for reporting.

Would you please apply the patch and let us know the behavior?

Meanwhile,We are working on a better solution if feasible.

What kind of query do you try to match ?

IMHO for many2one we can have:
(field, 'in', [])
(field, 'not in', [])
(field, 'in', [int, int, ...])
(field, 'not in', [int, int, ...])
(field, 'in', ['char', 'char', ...])
(field, 'not in', ['char', 'char', ...])
    a tuple can replace a list

(field, '=', 'char')
(field, '!=', 'char')
(field, '=', int)
(field, '!=', int)
(field, '=', bool)
(field, '!=', bool)
(field, 'child_of', 'char')
(field, 'child_of', [int, int, int])
(field, like_op, 'char')
    with like_op: ('like', 'ilike', 'not like', '=like')

Any other kind of query is an alien and must produce an error:
(field, 'not in', 'char') is wrong
(field, '=', ['char', 'char']) is wrong

Are you agree with me ? If not please correct me.

Hello JFB,

      I agree with you that (field, '=', ['char', 'char']) is wrong but (field, 'not in', 'char']) is not wrong as the code is made like that to support it. 'not in' gets replaced by '!='. Still few changes were necessary that are made and I have attached a solution for that. Would you please check the attached patch and notify us?

Thanks

> (field, 'not in', 'char') is not wrong as the code is made like that

Is it really necessary to complexify the code for unused and unlogical cases ?

Please keep it simple and bug free: remove thoses many2one queries in your code:
  [(field, '>=', ...)], [(field, 'not in', 'char')].

Hello JFB,

We cover as many conditions and not letting any error occur.

Can you please check with the latest patch?

Thanks.

Fixed in : by

Stable : 2138 <email address hidden>
Trunk : 2831 <email address hidden>

Thanks.

Hello senthilnathang,

It would sound better if you specify reason why you performed change in 'state' of this thread?

I hope you are aware about the meaning of each state.

Thanks.