Page 1 of 1

Meaning of "command sign differs"

Posted: Thu Feb 11, 2010 3:22 pm
by jzwanzig
Hi all,
my public branch fails one test, tv6_43, on the build farm with the failure information

Case_43
The diff analysis cannot be pursued : the command sign differ.
File 1, line 895, 89 ignored, character:
File 2, line 895, 89 ignored, character:=

Summary Case_43 : fatal error, see above message

This test runs fine on my machine. On my machine line 895 of the output file is a string of "=" signs beginning in the left most column, I assume this is the problem? But I'm not at all sure what I should do about it (my code additions didn't add any output writing). Any help you can suggest would be much appreciated.

thanks,
Joe

Re: Meaning of "command sign differs"

Posted: Thu Feb 11, 2010 5:45 pm
by gonze
Hi Joe,

From the information provided on the buildbot Web pages, you could not identify what was the problem.
You should have had access to the full t43.diff file to see what was going wrong. It is planned
to give full access to these *.diff files in the future (and no more the 50 last lines).

Now, back to your problem : I did access the testf machine, to see what was wrong.
It is only a very small problem of portability. Actually, using nstep=10 should fix it.
If you agree, I can fix it and proceed with the usual week-end merge with your branch included ...
Indeed, this is the only problem of your branch.

Best regards,
Xavier

Re: Meaning of "command sign differs"

Posted: Thu Feb 11, 2010 5:57 pm
by jzwanzig
Hi,
thanks for the tip. Are the test files always supposed to have an nstep=10 limit? I can add that and push again.

Re: Meaning of "command sign differs"

Posted: Thu Feb 11, 2010 8:59 pm
by mverstra
I agree - the error message is horrible - I have changed it in my 6.1.0

Matthieu