Script interpreter path length limit (POUNDBANGLIMIT)

Script interpreter path length limit (POUNDBANGLIMIT)

[Dennis Sperlich]

Hi,

i noticed an strange effect when calling a (python) script where the
interpreter is specified in the first line as follows:
#!/cvmfs/atlas.cern.ch/repo/ATLASLocalRootBase/x86_64/python/2.7.4-x86_64-slc6-gcc48/sw/lcg/external/Python/2.7.4/x86_64-slc6-gcc48-opt/bin/python

When i call this script from within zsh i get the following error:
zsh: $HOME/.local/bin/pip: bad interpreter:
/cvmfs/atlas.cern.ch/repo/ATLASLocalRootBase/x86_64/python/2.7: no such
file or directory

It seems like there is a arbitrary 63 character limit on the interpreter
char buffer. At least this is what i understand from the code in
Src/exec.c starting line 431.

So i'm wondering whether this is intentional and i'm not supposed to
point to interpreters with longer paths or just there due to historic
reasons? (i briefly checked, bash seems to have a limit of 79 characters
:-) )

Cheers
Dennis

Re: Script interpreter path length limit (POUNDBANGLIMIT)

[Bart Schaefer]

On Aug 10,  8:41pm, Dennis Sperlich wrote:
}
} It seems like there is a arbitrary 63 character limit on the interpreter
} char buffer. At least this is what i understand from the code in
} Src/exec.c starting line 431.

The specific choice of 64 for the buffer size (63+nul) is historical and
loosely based on OS limitations of the 1990s.

A LOT of info about this here:

    http://www.in-ulm.de/~mascheck/various/shebang/#results

Rather lacking in very-recent OS's, but you get the idea how inconsistent
this is.  I can't find anything about a standard way to look up the limit
on the build architecture, probably because POSIX defines #! as optional.

The limit in zsh could be expanded to allow the OS to flag the error.