On 14/10/10 13:50 +0200, Bertrand Chenal wrote: > Le Thu, 14 Oct 2010 12:14:37 +0200, > Cédric Krier <cedric.kr...@b2ck.com> a écrit : > > > On 13/10/10 17:26 +0200, Udo Spallek wrote: > > > Hello, > > > > > > please review my first Tryton test cases using the Proteus client > > > library at: http://codereview.appspot.com/2478041 > > > > I made a rewrite of this one using doctest to show the difference. > > > > http://codereview.appspot.com/2507041 > > > > IMO the "def test00XX" in the Udo version are not really needed and > obviously just add noise, without it and with some extra comments it > would be very close to the doctest version. > > Also, I think we will need tools like assertRaises, assertDictEqual, > etc (see [1]). They are not needed in the current example because those > are not finished: they create records but they do not test that results > are valid (like accounts balance, invoice amounts, etc).
You are still mixing unit testing and scenario concept.
> The problem is that those assert* method cannot be used outside
> unittest.testCase instances. OTOH relying on the doctest to validate
> result is also a problem:
>
> this validates:
>
> """
> >>> x = {1:1, 2:2}
> >>> x
> {1: 1, 2: 2}
> """
>
> but this does not
>
> """
> >>> x = {2:2, 2:2}
> >>> x
> {2: 2, 1: 1}
> """
> (wrong order of keys when dict is printed).
>
>
> and this does not too:
>
> """
> >>> x = {1:1, 2:2}
> >>> x
> {1:1, 2:2}
> """
> (missing space after ":").
The doc says you must test like this:
>>> x = {1:1, 2:2}
>>> x == {2:2, 1:1}
True
> Another problem of doctest is that it force you to write python code
> inside strings which reminds me bad PHP scripts.
This is because your mind see it like that.
You can see the all as a new language (a little bit like WEB [1]).
> Summary: We need assert* methods from unittest but unnitest layout of
> tests is not good. Doctest validation is over-sensitive on
> non-deterministic details and IMO best fit smaller tests.
There is some trick when testing some kind of values but it is really not
complicate to bypass.
[1] http://en.wikipedia.org/wiki/WEB
--
Cédric Krier
B2CK SPRL
Rue de Rotterdam, 4
4000 Liège
Belgium
Tel: +32 472 54 46 59
Email/Jabber: cedric.kr...@b2ck.com
Website: http://www.b2ck.com/
pgpQmqgP6qJTj.pgp
Description: PGP signature
