Re: [PATCH] tests: check token numbers.

[Akim Demaille]

Le 14 mai 09 à 09:54, Joel E. Denny a écrit :

> On Tue, 5 May 2009, Akim Demaille wrote:
>
> >         * tests/input.at (Numbered tokens): New.
>
> On branch-2.5, this test case fails for me as follows:
>
> 17. input.at:681: testing ...
> ./input.at:693: VALGRIND_OPTS="$VALGRIND_OPTS --leak-check=summary -- 
> show-reachable=no"; export VALGRIND_OPTS; bison input.y
> --- -   2009-05-14 00:07:35.868752224 -0400
> +++ /home/jdenny/cs/bison/bison-git/tests/testsuite.dir/at-groups/17/ 
> stderr	2009-05-14 00:07:35.000000000 -0400
> @@ -1,3 +1,3 @@
> -input.y:12.12-20: tokens HEXADECIMAL_2 and DECIMAL_2 both assigned  
> number 16702650
> +input.y:11.8-20: tokens DECIMAL_2 and HEXADECIMAL_2 both assigned  
> number 16702650
> input.y:9.8-20: tokens DECIMAL_1 and HEXADECIMAL_1 both assigned  
> number 11259375
>
> On master, it fails as follows:
>
> 17. input.at:676: testing ...
> ./input.at:688: VALGRIND_OPTS="$VALGRIND_OPTS --leak-check=summary -- 
> show-reachable=no"; export VALGRIND_OPTS; bison input.y
> --- -   2009-05-14 00:18:29.854049565 -0400
> +++ /home/jdenny/cs/bison/bison-git/tests/testsuite.dir/at-groups/17/ 
> stderr	2009-05-14 00:18:29.000000000 -0400
> @@ -1,3 +1,3 @@
> -input.y:12.12-20: tokens HEXADECIMAL_2 and DECIMAL_2 both assigned  
> number 16702650
> -input.y:9.8-20: tokens DECIMAL_1 and HEXADECIMAL_1 both assigned  
> number 11259375
> +input.y:10.12-20: tokens HEXADECIMAL_1 and DECIMAL_1 both assigned  
> number 11259375
> +input.y:11.8-20: tokens DECIMAL_2 and HEXADECIMAL_2 both assigned  
> number 16702650
>
> It looks like the hash iteration order is undefined.

Yes, indeed.

> I suppose the test
> could check for each possible order, or is there a better way to  
> handle
> this?

Instead of relying on the collision to check the hexadecimal support,  
I can simply check the numbers in the output files.  Using the  
collision was the shortest path, but I did not think about this issue.

I'll see if I can find something better, thanks for pointing this out!