db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From TomohitoNakayama <tomon...@basil.ocn.ne.jp>
Subject Re: DERBY-273, owner Tomohito Nakayama, Is it time to move this test out of the nightly regression suite?
Date Fri, 10 Feb 2006 16:44:50 GMT

About this issue, I think we should be very careful about removing any 
sign of problem...

My intention is to add test implemented by DERBY-511, compensating for 
ignoring shutting down exception in DERBY-273.

My position was watching response of other persons for DERBY-511  ...
Well ... There were almost no response .

If there were no response in days,
I will commit DERBY-511 and
alter standard input/output where ignorable shutting down exception was 
printed, as discussed in DERBY-273,
under the concept of lazy vote.

// Comments:
// I have some doubt in just removing output of errors ,which we can 
ignore,  just aiming the goal.
// I feel this kind of goal is just goal for goaling, and not the goal 
to reduce the problem.
// I think purpose of testing is *not* getting the result of "no error" ,
// but to try running program in many circumstances to find and fix 
// Therefore, in some cases, it may be correct to pass the result of 
testing which have some ignorable errors.
// If we can ignore them , we should just ignore them and no need to remove.
// Though the ignorable errors may be annoying, I think these errors may 
tell us some failure in unknown circumstance ...
// It would be just matter of time and effort for handling errors.
// Well. This would be *my* persistence and not others thought. And I 
don't think this my opinion is absolute.
// I won't veto others, who believe it is no problem ,to remove 
ignorable errors.
// // It may be also truth that ignorable errors makes the result 
// // However, I noticed that it would be the right answer , to write 
the information about ignorable errors in Wiki and ignore them,
// // instead of removing them. How do others think ... ?

Best regards.

Kathey Marsden wrote:

>Mike Matrigali wrote:
>This issue is related to  how Network Server should  handle exceptions 
>which occur closing down connections if the database has been
>intentionally shutdown.  Currently they print to the console (System.out
>by default)
>One option to keep the test running cleanly while that is being worked
>out is for the test to specify an alternate console than System.out. 
>That I think is a good solution for this test in the short term, but I
>recall that Naka had some concerns about that.


        Tomohito Nakayama



View raw message