Re: dealing with executable shell scripts

[Andrew W. Nosenko]

On Tue, Mar 20, 2012 at 15:47, Nick Bowler <address@hidden> wrote:
> On 2012-03-20 07:00 +0900, Miles Bader wrote:
>> Is there a recommended way for dealing with binaries that are simple
>> shell scripts in automake?  I currently use something like the
>> following:
>>
>>    bin_PROGRAMS = myprog
>>
>>    myprog_SOURCES = myprog.sh
>>    myprog: myprog.sh
>>
>>    %: %.sh
>>            $(shbin_verbose)cp $< $@; chmod +x $@
>
> Notwithstanding what others have already said, the above rule has some
> issues of its own:
>
>  (1) If the "chmod" command fails: the failure will be noticed (make
>     will stop), but the target file has already been created at this
>     point.  The target will thus be considered up-to-date on a
>     subsequent make run, despite the fact that it was not created
>     correctly.
>
>  (2) If the "cp" command fails: the chmod is still attempted, which may
>     succeed if the target file exists but is out-of-date.  If chmod
>     succeeds, the failure will not be noticed by make, and the build
>     will continue using an out-of-date target.
>
> Moreover, the use of GNU make-specific pattern rules here is pointless
> because the same can be accomplished by a more portable single-suffix
> rule.
>
> Better to write it as follows:
>
>  SUFFIXES = .sh
>
>  .sh:
>        $(shbin_verbose) cp $< address@hidden
>        $(AM_V_at) chmod +x address@hidden
>        $(AM_V_at) mv -f address@hidden $@
>

Why do not use install(1) instead of cp(1)+chmod(1) or cp(1)+chmod(1)+mv(1)?

-- 
Andrew W. Nosenko <address@hidden>