axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Davanum Srinivas" <>
Subject Re: [Axis2]Running build before commiting code
Date Sat, 04 Nov 2006 00:03:27 GMT

There's a clover target in maven that used to work. Please help us by
running that to identify holes in our test coverage.


On 11/3/06, Alan M. Feldstein <> wrote:
>  Eran Chinthaka wrote:
>    If we exclude something from the tests, then no one will run it,
> whatever you say. This is what happened to itests in integration folder.
> I for one didn't run that for some time, did you?
> So what I suggest is to keep the default behavior as is and if some one
> knows what he is doing then he needs to put the flag , Ruchith put
> ,explicitly.
> See, we can not afford to break anything at this moment. I encounter
> couple of bugs which happens to happen again-and-again. I fixed it all
> the time but some one breaks it again. The reason is lack of tests.
> Every test in integration module or any other module makes sure
> particular features are working properly.
> I prefer adding more tests than removing it. Yes it is pain when you do
> a build, but it is for the betterment of long term stability of Axis2.
> Please use the proper flags sensibly.
>   Eran is absolutely right.
>  When I fixed some bugs in UnsignedLong, I was amazed to find that none of the test cases
covered the upper half of the number space! (I added one.) Those are the bit patterns that
a primitive long would consider to be negative, but an UnsignedLong would be required to interpret
as large positive numbers. In other words, the most important test cases for the UnsignedLong
class were missing.
>  If this is representative of test coverage in Axis2, then we have a problem. If anything,
more testing is required.
> --
> Alan Feldstein

Davanum Srinivas : (Oxygen for Web Service Developers)

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message