[BRAINSTORM] shebang fixing framework

[Chris Rees]

On 30 Apr 2013 18:46, "Eitan Adler" <lists at eitanadler.com> wrote:
>
> On 30 April 2013 13:41, Baptiste Daroussin <bapt at freebsd.org> wrote:
> > On Tue, Apr 30, 2013 at 01:38:03PM -0400, Eitan Adler wrote:
> >> On 30 April 2013 13:35, Baptiste Daroussin <bapt at freebsd.org> wrote:
> >> > All of this will be damn slow operations, we want to keep this as
fast as
> >> > possible, even if per-ports it seems quite fast, when you are about
the build
> >> > the whole ports tree few seconds there and there will become hours
> >>
> >> Yes, that is why I ended with "I guess it isn't too bad.", although I
> >> guess it wasn't clear I meant needing lang_FILES.
> >>
> >> I'd prefer shebangfix_FILES and have all the language fixes apply
> >> without needing to specify which one.
> >
> > most of the case you will know what file needs what shebang, so why
trying all
> > possible case when you can get it directly?
>
> If applied as a regex it doesn't really matter.
>
> I'm thinking it will look something like
>
> sed -e regex_for_perl \
>     -e regex_for_python \
>     -e regex_for_php \
>     ${shebangfix_FILES}
>
> or some variant thereof
>
> now that I think about it a bit more it may even be faster than
> forking once per language.

How many ports are likely to have more than one language?

This proposes running each regex over each file in turn.

Chris