| Newsgroups |
gmane.comp.lang.ocaml.beginners |
| Message-ID |
<CAMu2m2KGRWkf-wa5_zm2M2pfGEYJ0PfDfCH0FDezipDnAHYaTg@mail.gmail.com> |
On Thu, Sep 17, 2015 at 8:45 PM, Oliver Bandel [email protected]
[ocaml_beginners] <[email protected]> wrote:
> - Jenga. It mostly takes the right approach.
>
> Why does it take "the right approach"? What criteria do you used for
> this statement?
>
It is a library, which I feel is the correct way to write essentially any
code. I never quite understand when a command line tool's functionality
isn't also provided as a library function. It provides some core constructs
representing what a build rule is, and builds higher level functionality on
top of that. It's OCaml, so I can write arbitrarily complex code in a
language I like. In other words, just the usual criteria for any software;
it makes easy things easy and hard things possible.
... except that it doesn't yet. Actually it's main build can only be
triggered by calling the jenga command line tool, which loads your jenga
config file. Thus, it already becomes quite hard to understand precisely
what is the input to a build, and what is the result. Making it a library
function would clarify that and ease debugging (I wanted to test out some
things from utop and it was rather difficult or impossible to do). Also,
the semantics of various functions are not clear to me, which is maybe just
a documentation issue or maybe a design problem.
> The tool is an OCaml library,
> > so you have the full power of OCaml to write your build system.
>
> Hmhh, wouldn't it be better to have something that is independent
> of OCaml itself (orthogonal, and possibly also working for other
> languages)?
>
OMake, ocamlbuild, and jenga all can be used for non-OCaml projects. In
terms of the implementation language, why choose another language. OCaml is
the language we like, and it is perfectly capable for implementing a build
system.