ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephane Bailliez <>
Subject RE: proposal/junit
Date Wed, 16 Jan 2002 11:40:59 GMT
> -----Original Message-----
> From: Peter Donald []

> Could you give us a rundown what you are doing in the 
> proposal dir. I see the 
> junit task has a remote socket based server thingie and it 
> looks interesting 
> ;)


Here are some points thrown with no particular order.

- The goal was to come up with a new design since it starts to be a real
mess since it is made of bunch of patches here added incrementally and the
current architecture is limited.

- The main point was to allow to run JUnit tests in isolation in another VM
and receive results in another so that we can have better control if
something hangs and possibly configure the VM with particular settings.
(encoding, memory, system properties, classpath, etc...)
The main problem in the actual design for example is that a failed test
(cancelled) would result in no report at all and no way to know that
something failed.

- I changed the design of the formatters to allow filtering of events (for
example the stacktrace filter). I would like to allow formatter
configuration so that for example we could configure a JMS formatter that
send events on a given topic/queue, or on a SocketFormatter, RMIFormatter

- The protocol used for socket communication is based on some code initially
done by Erich Gamma in the JUnit Eclipse plugin, I extended it a little.

- It is able to collect tests on a one-by-one basis or in a batch way from
directories or archives (awaited feature apparently).

- It is *totally* incompatible w/ the current task.
For example I do not assign a formatter to a test or a batchtest since it
has no real sense to me, they are global to JUnit. I can't see a needed use
case of assigning different formatters to differents tests. If it is needed
then run different JUnit to have different reports.

- It is not working yet. (the task configuration, otherwise the runner seems
to work fine)


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

View raw message