[CMake-Promote] simplified vs. fullblown languages

[Brandon Van Every]

On Jan 31, 2008 2:00 PM, Brandon Van Every <bvanevery at gmail.com> wrote:
>
> Anyways, in the course of events it was mentioned that someone ditched
> CMake in favor of SCons because of CMake's custom language.

And then they moved on to Premake.  Turns out this was for a ~40K loc
small-to-medium sized project.  The pattern I'm seeing is that things
like Premake are getting the design wins for small projects.  I don't
know whether it can scale up or what the largest Premake project is.
I don't know if Premake offers compelling advantages for small
projects, or if it's just perception along the lines of "I already
know and like Lua or Python or whatever, so I want something that uses
what I know."  That would make sense for a small project, because
people are probably not feeling terribly committed to learning
anything about build systems in order to keep their small projects
going.  That's what happened to me with Chicken Scheme, I think.

> Marketing-wise, I think there's a fork in the road here that CMake
> needs to address.  Many programmers believe they want a "full
> featured" programming language, and we should deal with the truth or
> falsehood of that perception somehow.

An observation: it's simpler to maintain code in a simple language, if
that simple language actually expresses the things you need to do.  I
think that a lot of programmers think they're going to need to
customize everything.  In fact, a significant contingent actually
wants to customize everything, because that's fun and profitable for
them personally, whether it's beneficial for the project or not.  If
CMake could actually demonstrate what it's capable of doing "off the
shelf," that could be weighted against the perceived need to
customize.  Against this, however, I think we need to be responsible
and forwards looking about what people actually need to customize.


Cheers,
Brandon Van Every