Re: FAIL: rev 5064

Philip Martin <philip@codematters.co.uk> writes:
> Should the commit fail? The test relies on a "Non-ascii" error
> message, that can only be generated in the code path that is run if
> UTF-8 conversion is unavailable.

Yes, that's wrong... I've gotten rid of that portion of the test for
now (see revision 5068).

Is there any portable way to keep the control-char test?

Oh, wait, yes -- test for *either* the error that is generated when
UTF-8 conversion is unavailable, or the error that is generated when
UTF-8 conversion failed. That is, if encounter one of those two
errors, then the test succeeds.

Does that sound reasonable? This test has proven so error prone that
I'd rather the idea were reviewed *before* I implement it :-).

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Feb 24 22:22:37 2003