On 7/9/06, Jim Gifford <[EMAIL PROTECTED]> wrote:
Dan,
I have to disagree with you on this. The testing should be done by
hand not by a script. That takes the educational value our and plus
gives a lot of false positives during a release cycle.
You have a point. I will prepend me request to say "If you plan on
scripting the build...". You're right that false postives can creep in
on scripts. Certainly someone should be doing the build by hand to
ensure that nothing is missed and that the printed commands work as
advertised.
However, I stand by the use of jhalfs for testing. If it's accurately
depicting the book, then there is no faster or more consistent way to
test it. Something I didn't mention as one of my motiviations is that
with wider testing, any bugs will be caught in jhalfs. I personally
feel that George and Manuel have stress tested the tool very well and
that any bugs left are purely from a usablitilty standpoint. I really
do believe that it's the ultimate testing tool if it's working
correctly, though.
So, I'll say that I hope that anyone that was going to test the build
with a script will use jhalfs as that script. And I hope that others
will use jhalfs to make it more robust as a testing tool.
--
Dan
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
