Re: [address@hidden: Re: Type-in programs using BASH]

[Ryan Cunningham]

First of all, it's 2015, not 1982. (By the way, I got my inspiration for this from the July 1984 issue of COMPUTE! Magazine, which I got from the OpenLibrary project of the Internet Archive.)

 

Second, the machine code we type in, if we receive any, will most likely be 8086 machine code, not Motorola 68000 machine code.

 

However, there are a lot of other Linux kernel--supported platforms out there: PowerPC, Intel Pentium, and SPARC are a few.

 

I am proposing this as a possible alternative or complement to publication on the Internet to take into account those without Internet access, though those with Internet access also get the benefit.

 

I'm CCing to the GCC mailing list. (For readers there, the rest of the thread is accessible at http://lists.gnu.org/archive/html/bug-bash/2015-01/threads.html.)

On Mon, Jan 26, 2015 at 1:45 PM, Greg Wooledge <wooledg@eeg.ccf.org> wrote:

On Mon, Jan 26, 2015 at 01:32:21PM -0800, Ryan Cunningham wrote:
> When you receive a program in *object code* form, you would type it into an
> object code editor and then save it in a binary file.

Why?  Is it 1982 again?  Are we typing in 6502 machine code from a
glossy magazine?

In the 21st century, if I "receive a program in object code form", it's
almost certainly going to be a digital delivery.  I'll have the program
in the form of bytes on some kind of machine-readable medium (floppy,
CD, DVD, USB mass storage device), or I'll be receiving it from a
network socket (HTTP server, FTP server, etc.) and can store it on
nonvolatile medium of my own (hard drive, USB device, etc.).

But, moving on....

> You would then run
> `chmod +x' on that binary file and then tell Bash to execute it (as
> *opposed* to interpreting a Bourne Shell script), hoping you didn't make a
> mistake.

Executing it doesn't really involve bash at all.  You might use bash
to tell the kernel to execute it, but it's ultimately the kernel that
performs the loading and execution.

> Again, I apologize for the confusion.

But... what's the point of this discussion?  Wasn't it something to do
with the GNU General Public License?  I only glanced at the earlier
messages, but that was my gist.

If you're trying to make some kind of legal argument about the GNU GPL
then you must be extremely precise.  Typing "./foo" in bash does NOT
constitute the construction of some kind of derivative work which would
invoke the constraints of the GPL, or any other license.  Typing "./foo"
does not load object code into bash's memory space.  It is NOT similar
to loading a shared library or a loadable builtin.

It doesn't matter where the object code came from or how it was placed
into your file system.

--

Ryan Cunningham