Re: testcase for "absolute source" patch

[Baurjan Ismagulov <ibr@ata.cs.hun.edu.tr>]

Hello Michael,

On Wed, Aug 18, 2004 at 03:00:06PM -0400, Michael Chastain wrote:
> The problem is default_target_compile in dejagnu.
> default_target_compile really wants to be in charge of downloading files
> from build to host and managing the filenames on the host.  This
> directly conflicts with your script, which wants to manage the names
> on the host.

As far as I could understand, there are also other tests that do not
work well in build != host scenario. I don't remember any decision from
the recent thread on this issue. OTOH, this test functions locally and
checks a number of cases, at least two of which have been subject to
regressions. I think this was the reason why Andrew asked for it in the
first place. While I agree that fixing the infrastructure is also
important, may I suggest that we have at least local checking in the
trunk? After all, whatever the state of the infrastructure, we still
want to check for bugs; a non-ideal test is better than no test at all,
IMHO. I also don't see any problem in modifying the test as the
infrastructure changes, whereas not having the test is going to lead to
bugs, especially given an open PR involving the respective subsystem.

To summarize: the test is working probably for 95% of the users. I'm
willing to make the test as close to the ideal in your head as possible,
and I can do it only now. I suggest that we improve what can be improved
and leave the rest for better times.

What do you think?

With kind regards,
Baurjan.