> > > I think POSIX would argue that -buildfile is the '-b' option
> > > with an argument of 'uildfile'.  Or possibly equiv to
> > > -b -u -i -l -d -f -i -l -e.
> > 
> > I would agree, except that Sun seem happy to use the -buildfile style.
> > Supporting both seems important to me.
> 
> That's a slipperly slope though.  We're all angered when Microsoft
> embraces and extends a technology or standard.  I think we should have
> a similar reaction when Sun, IBM or anyone else does it too.
Agreed.  

> It'd be fine, if we follow John's suggestion of having different
> styles of parser, possibly even with the POSIX one as the default.
> You should have to go out of your way to do something non-standard
> (non-POSIX, in this case).
Cool.  When I have the multiple value patch together I was thinking of
having a bash at refactoring some of the parsing code anyway (its just
a bit hard to follow at the moment, and even more so with the multiple
value stuff going in).  So I will have a go at drafting a design for
this.  I think the main thing here is to have it as simple as possible.
The last thing we want is for users to wonder "what the hell is going 
on, all I want to do is parse a command line?".

An idea I have for testing is to add an ApplicationTest unit test that
has testXXX methods, where XXX is an real world command line utility
e.g. testJava, testJavac, testTar, testChmod, testUname, testCd, etc.
Not only will we have some good test coverage we'll also have ready
to wear demos :-)

-John K


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to