Kent and Mathieu,
Sorry about the lack of test and the buggy code I suggested. I only meant=
suggest an improvement to Junit, for a problem that caused me trouble. On=
local settings, I have overriden those two Assert methods and thought tha=
it might be helpful to other Junit users. I thought that as a user of
open-source software, I should spent time to try to contribute possible
improvement to those packages.
However, I found it difficult to contribute suggestions to Junit for the
- the Junit homepage http://junit.sourceforge.net/ does not provide links=
the mailing lists. One needs to find the Junit sourceforge project summar=
to find this (http://sourceforge.net/projects/junit/)
- on the sourceforge project summary, it is difficult to know where to po=
contributions: the forums seem partially dead
http://sourceforge.net/forum/?group_id=3D15278, and it seems to be the sa=
thing for the junit-users mailing list
(http://sourceforge.net/mail/?group_id=3D15278) In doubt I've posted to t=
junit-devel mailing list (albeit rapidely consulting the archives I mostl=
find automatically generated mails). I also had a hard time finding the
Yahoo egroups list http://groups.yahoo.com/group/junit/. Mentionning it o=
the project homepage would certainly help.
Please, take my remarks positively. I am very grateful to the team of Jun=
developers and I think the large and constantly increasing community of
Junit users is thinking the same way. Because I find Junit useful to my w=
and I believe it helps the software community improving (I mean in
comparison to more mature engineering communities like mechanic or
electronic engineering), I try to contribute little things when I can. I
understand this is how open-source projects succeed...
Concerning extensions to Junit, I wonder if Junit could be suited to perf=
larger scale tests. In particular, testing at the level of components (sa=
component which implements the JMS API). My project is broken up into
components exposing interfaces. In addition to unit tests, I find it
important that inter-component interfaces to act as a kind of contract. A=
such, I find it valuable to be able to automate testing of such contracts.
For what I understand about Junit, it was not originally designed for thi=
granularity of tests. Do you think, that it should support this kind of
tests, or would it better be a separate focused project? Is there such
projects already available?
In the same idea, I started a discussion on the Yahoo egroups list
http://groups.yahoo.com/group/junit/message/3841 concerning the support t=
test non-standard distributed application (by non-standard I mean
distributed applications not using HTTP or EJB containers as a
communication/distribution infrastructure). In my case, I need to test
components which support inter-process communication (such as a JMS
implementation), and was planning on using Junit to do this. After
collecting feedback and not finding an existing infrastructure, I
implemented a first version of a remotized version of some Junit classes
(TestListener, Test) and some support to spawn test case on remote JVMs.
Would this be of interest to the Junit project as a base for an extension
supporting non-standard distributed applications?
Thanks again for your work on Junit, and sorry for the long mail :-)
> -----Message d'origine-----
> De : Kent Beck
> Envoy=E9 : mercredi 23 janvier 2002 19:44
> =C0 : Guillaume Berche
> Objet : RE: [Junit-devel] CONTRIB: fix suggestion for
> What are the test cases? The implementation we can figure out :-)