Re: [Etester-users] Mailing list usage

[Gunnar G. <gu...@cs...>]

Hi All,

At 11:59 PM -0400 7/17/03, Dave Makalsky wrote:
>Hi:
>
>Please encourage 3311 students to join the etester-users list by 
>going to http://lists.sourceforge.net/lists/listinfo/etester-users
>
>We will save the etester-devel list for development issues.

I do not suggest such encouragement.  I do not relish receiving 
messages from an ever broadening list.   I suggest we could encourage 
them to send to the developers of etester who can maintain a list of 
problems without everyone on the list receiving every complaint.

In fact, I am unsubscribing  from the list for two reasons.  (1) I am 
going on holiday and do not want to come back to a flood of messages 
about etester.  (2) I am not that interested in etester that I want 
to devote significant time to it.

My interest in etester lies only so far as it is it may be of use in 
the software design course.     I am not convinced that the overhead 
of using it is worth the effort, either for myself or for students. 
I have used etester in the dictionary exercise (W03 course) and did a 
minimal output test program.   The major difference I see is a colour 
html output screen but I see no other advantages/benefits.  I do see 
disadvantages.

(0) with minimal output testing I can do incremental development and 
frequent testing, even test driven design.  The java libraries I 
point to from my previous www pages for the course were constructed 
incrementally with the test programs growing as the libraries grew.

(1) etester does not give me additional information that I do not 
have from a minimal output test program.  In fact, with a minimal 
output test program, I am in control of what and how much is 
reported.  Instead of being restricted to what other people think is 
good or useful.    For example, the discussion as to where comments 
should go, there is no right or wrong.  There are only opinions and 
styles of programming.  Another example is the poor commenting 
facility in Eiffel that makes it difficult to write longer comments 
forcing people to the style that Myer prefers but has many drawbacks 
as one-liners are not sufficient documentation for many features. 
If block comments (like C /* ... */) were implemented, then people 
who prefer short comments can have them and people who want long 
comments can have them.

(2) When I compare my programming effort with constructing a set of 
minimal output tests there is little difference with etester.

(3) minimal output testing can be used on any platform for any 
programming language without the use of special tools created by 
other people.  People can carry their style over to any project.  Now 
there is one style for etester, one for java,  potentially one for 
each programming language/environment.

(4) There is a significant cost in using such tools ranging from 
installation to understanding and using the tool.  That overhead has 
to be compensated with information that could not be easily obtained 
otherwise and/or significantly reduced programming effort.   I have 
not found such compensation with etester.

(5) To me it looks like more research into the construction and use 
of etester is needed before it will of significant use for students.

Cheers ... Gunnar