FWIW, I agree with Ian. I believe that NAnt should be a declarative tool and not permit procedural style code to be written at all. If the extra (complex) functionality is there then it will be used, and NAnt will get a reputation for being unwieldy.
Just my opinion Mark > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED]]On Behalf Of Ian > MacLean > Sent: 28 August 2002 01:50 > To: Shaw, Gerry > Cc: [EMAIL PROTECTED] > Subject: Re: [nant-dev] Property names and functions > > > > > > >I'm starting work on adding functions to nant and the syntax I was > >thinking of using is this: > > > > > > > Are we sure we need this. No disrespect but couldn't we do a lot of > whats in these functions using the script task ?. Its a slipperly slope > to be taking - just look at how xslt has evolved from having a few > functions thru script and now a full extension model. The real questions > is "Do we want NAnt to become a programming language ? " and I'm not > sure we do for a couple of reasons. 1) simplicity has been a stated goal > of the project and adding a whole bunch of programming language > constructs to a declarative build tool is certainly going to complicate > things. 2) writing procedural code in xml isn't pretty. Just look at an > xslt example > <xsl:choose> > <xsl:when test="$sortby='scientific_name'"> > <xsl:apply-templates select="//raptor"> > </xsl:apply-templates> > </xsl:when> > <xsl:when test="$sortby='status'"> > <xsl:apply-templates select="//raptor"> > </xsl:apply-templates> > </xsl:when> > <xsl:otherwise> > <xsl:apply-templates select="//raptor"> > </xsl:apply-templates> > </xsl:otherwise> > </xsl:choose> > > compared to > switch(condition) { > case "A": > DoA(); > break; > case "B": > DoB(); > break; > case "PerlManagedExeWiz": > DoC(); > break; > } > in C#. I like being able to do declarative stuff in NAnt and procedural > programming in a fully fledged programming language. As soon as I start > doing say string manipulation in xslt I get enormously frustrated and > want to use a real programming language that has a proper string > library. It looks like the initial functions you've proposed are quite > similar to those in in xslt which leads me to think that doing > programming in NAntScript would be just as frustrating. > > 3) the Ant team has had similar discussions and has always resisted > going too far down this route - maybe Stefan can give us more reasons > for why if he's reading. > > I'm not trying to be unduly negative I just thingk we should think very > carefully before turning NAnt into a programming language. > > my 2c > > Ian > ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Nant-developers mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/nant-developers
