ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Clark Archer <clark.arc...@gmail.com>
Subject Re: Introduction and possible new feature for Ant
Date Sat, 13 Mar 2010 15:29:53 GMT
On Tue, Mar 9, 2010 at 11:19 PM, Stefan Bodewig <bodewig@apache.org> wrote:

> Hi Clark
>
> On 2010-03-09, Clark Archer <clark.archer@gmail.com> wrote:
>
> coded up a HistoryRecorder which builds a simple text file containing a
> > history of up to the last 100 executions for each test class/suite.  I
> track
> > pass/fail and execution time for each test class and update the history
> file
> > at the end of each <junit /> task.
>
> This sounds interesting.  Despite all the hate XML gets it may be a more
> suitable format, though.  You could add more structured data and create
> all sorts of reports later.
>
> What would a snippet of such a file look like?
>

Here are a few lines of one of the history files:

com.nprise.datalist.ColumnMapperImplTest,pass,15,pass,6,pass,8,pass,4,pass,3,pass,16
com.nprise.datalist.DataListBuilderSqlTest,pass,13,pass,33,pass,15,pass,16,pass,31,pass,36
com.nprise.datalist.DataListJoinerTest,pass,8,pass,8,pass,10,pass,9,pass,9,pass,30
com.nprise.datalist.DataListNonUniqueIndexTest,pass,23,pass,25,pass,22,pass,19,pass,21,pass,190

I don't have anything against XML (in particular), but I went with a text
file because I only had a little time to work on it last weekend and I don't
know the various XML APIs well.  Definitely XML would be more flexible for
adding more attributes later.  This file has to be read at the beginning of
each test run (or batch) and then updated at the end of the run.


> > I'm interested in getting some feedback on this feature as well as
> discuss
> > some of the design decisions regarding the overloading of the nested
> <format
> > /> attribute.   I discovered that the taskEnded life-cycle method of
> > ProjectComponent is not called when the tests are running from
> > JUnitTestRunner in a forked VM and came up with a solution (add more
> > life-cycle methods and avoid implementing ProjectComponent).  Not perfect
> > but it enables "recorders" to work correctly in a forked VM.
>
> Could you expand on this a little - maybe in a separate thread?  It
> sounds like a bug.
>
>
Definitely, I'll start a new thread about this.

-Clark A.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message