Re: #!/usr/bin/env tclsh

[Bart Veer <bartv@ecoscentric.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3466 bytes --]

>>>>> "Gary" == Gary Thomas <gary@mlbassoc.com> writes:

    Gary> Bart Veer wrote:
    >>>>>>> "John" == John Dallaway <john@dallaway.org.uk> writes:
    >> 
    John> This patch simplifies the #! magic used to invoke Tcl
    John> scripts by using "/usr/bin/env tclsh" to find the tclsh
    John> executable. Very old Cygwin installations providing only
    John> tclsh83.exe or cygtclsh80.exe are no-longer supported.
    John> Checked-in.
    >> 
    >> Actually, this patch has broken things in various ways. Consider e.g.
    >> file2c.tcl in the romfs package. The CDL invokes this using e.g.:
    >> 
    >> sh file2c.tcl testromfs_le.bin testromfs_le.h
    >> 
    >> With the old magic this still worked fine because sh would ignore the
    >> #! at the start completely and move on to the 'exec sh -c' on line 3.
    >> With the new '#!/usr/bin/env tclsh' the sh invocation ignores the
    >> #! comment on line 1 so ends up trying to run the whole Tcl script as
    >> a shell script. Needless to say this is not very successful.
    >> 
    >> io/framebuf is similarly affected. services/memalloc/common is not. I
    >> have not yet checked all the other packages that use Tcl scripts.
    >> 
    >> Possible solutions are:
    >> 
    >> 1) revert the change
    >> 2) remove the 'sh' bits from the relevant CDL scripts, treating the
    >> Tcl script as plain executables.
    >> 3) make the CDL invoke /usr/bin/env tclsh directly, treating the
    >> Tcl scripts as Tcl scripts.
    >> 
    >> (1) would be a bad move. I think I would prefer (3) to (2).

    Gary> Why isn't this working? According to 'man sh' on my Linux
    Gary> system:
    Gary>        If the program is a file beginning with #!, the
    Gary>        remainder of the first line specifies an interpreter
    Gary>        for the program. The shell executes the specified
    Gary>        interpreter on operating systems that do not han- dle
    Gary>        this executable format themselves. The arguments to
    Gary>        the interpreter consist of a single optional argument
    Gary>        following the interpreter name on the first line of
    Gary>        the program, followed by the name of the program,
    Gary>        followed by the command arguments, if any.

    Gary> It would seem that since Linux *does* handle this directly, 'sh'
    Gary> chooses to ignore it :-(

If you run 'sh xx' instead of just 'xx', is 'xx' still a  "program" by
the above definition is or is it just data for sh? I don't know the
official answer, but it appears that Linux sh treats it as just data.
    
    Gary> In any case, I vote for (2), otherwise you may end up with
    Gary> the same problem all of this was trying to fix in the first
    Gary> place, namely not knowing where/how to find 'tclsh'

Hence the "/usr/bin/env tclsh" in the CDL script as opposed to just
"tclsh". Although come to think of it, I am not sure that gains
anything. Either make (or the shell invoked by make) or the env
utility will end up searching PATH for tclsh.

Bart

-- 
Bart Veer                                   eCos Configuration Architect
eCosCentric Limited    The eCos experts      http://www.ecoscentric.com/
Barnwell House, Barnwell Drive, Cambridge, UK.      Tel: +44 1223 245571
Registered in England and Wales: Reg No 4422071.
Besuchen Sie uns vom 3.-5.03.09 auf der Embedded World 2009, Stand 11-300
Visit us at Embedded World 2009, Nürnberg, Germany, 3-5 Mar, Stand 11-300