Re: handle #! /usr/bin/env

[matthew sporleder <msporleder%gmail.com@localhost>]

On Mon, Mar 4, 2013 at 12:50 PM, Alan Barrett <apb%cequrux.com@localhost> wrote:
> On Sun, 03 Mar 2013, matthew sporleder wrote:
>>
>> Okay, sorry to develop this on the mailing list but I got a new update.
>> I'm mainly interested in comments about the idea.  I think it saves a
>> lot of hassle.
>>
>> 1{ /env -i/!s|^#![[:space:]]*/usr/bin/env|#!|;}
>
>
> You could do it all in a single command.  The idea is:
>
> If this is line 1:
>    If line has "#!", optional space, optional {"/use/bin/env"
>         followed by one or more spaces}, and finally the
>         interpreter you were looking for, then replace it all
>         with "#!" followed by the new interpreter.
>
> I don't see the need for a special case for "/usr/bin/env -i". Firstly, many
> systems (including NetBSD) prohibit multiple arguments on the "#!" line, so
> stuff like "#!/usr/bin/env -i foo" will fail often enough that people learn
> not to write such code.  Secondly, scripts in which you don't want the
> substitution to occur should not be passed to the REPLACE infrastructure.
> If, despite this, you want to avoid accidentally matching "/usr/bin/env -i
> foo", then you can change the REPLACE.sys-foo.old regexps to use
> "[^[:space:]]*" instead of ".*".
>
> Here's patch (untested).
>
> --apb (Alan Barrett)
>
> Index: mk/configure/replace-interpreter.mk
> ===================================================================
> --- mk/configure/replace-interpreter.mk 6 Feb 2012 06:18:36 -0000       1.11
> +++ mk/configure/replace-interpreter.mk 4 Mar 2013 16:19:09 -0000
> @@ -90,6 +90,9 @@ REPLACE.sys-sh.new=   ${SH}
>  REPLACE_FILES.sys-sh=  ${REPLACE_SH}
>  .endif
>  +# sed regexp to match optional "/usr/bin/env" followed by one or more
> spaces
> +REPLACE.optional-env-space=    (/usr/bin/env[[:space:]][[:space:]]*)?
> +
>  .PHONY: replace-interpreter
>  replace-interpreter:
>  .for _lang_ in ${REPLACE_INTERPRETER}
> @@ -99,7 +102,7 @@ replace-interpreter:
>
>         cd ${WRKSRC};                                                   \
>         for f in ${REPLACE_FILES.${_lang_}}; do                         \
>                 if [ -f "$${f}" ]; then                                 \
> -                       ${SED} -e
> '1s|^#![[:space:]]*${REPLACE.${_lang_}.old}|#!${REPLACE.${_lang_}.new}|' \
> +                       ${SED} -e
> '1s|^#![[:space:]]*${REPLACE.optional-env-space}${REPLACE.${_lang_}.old}|#!${REPLACE.${_lang_}.new}|'
> \
>
>                         < "$${f}" > "$${f}.new";                        \
>                         if [ -x "$${f}" ]; then                         \
>                                 ${CHMOD} a+x "$${f}.new";               \


Looks good to me.  Who can give a final vote on if this is a good idea or not?