[2747] in SIPB_Linux_Development
Re: Linux split in infoagents
daemon@ATHENA.MIT.EDU (Aaron M. Ucko)
Tue Jun 29 13:15:41 1999
To: "t. belton" <tbelton@MIT.EDU>
Cc: web-agents@MIT.EDU, linux-dev@MIT.EDU
From: amu@MIT.EDU (Aaron M. Ucko)
Date: 29 Jun 1999 13:15:15 -0400
In-Reply-To: "t. belton"'s message of "Tue, 29 Jun 1999 12:40:33 -0400 (EDT)"
"t. belton" <tbelton@MIT.EDU> writes:
> I'm currently trying to figure out the least messy way to split the
> platforms in infoagents. My problem is that my predecessors dropped all
> Linux things in a single place - all three i386_linux? platforms point to
> the same subdirectory tree. I KNOW, though, that some of the things in
> there are linux2 only, and a few (like the current version of 4.61 there)
> may even be linux3 only. I don't know how to sort them out.
Here's one way to determine what platforms SOMEBINARY runs on,
assuming you're logged into a linux3 box (Emil gave you access to
keesh, right?):
First, run
file SOMEBINARY
If the result is something like
SOMEBINARY: Linux/i386 demand-paged executable (ZMAGIC), stripped
then that means you're looking at an old i386_linux1 binary, which
will also work on linux2 and linux3 systems with appropriate
compatibility libraries installed.
On the other hand, if the result is something like
SOMEBINARY: ELF 32-bit LSB executable, Intel 80386, version 1, dynamically linked, stripped
then you're looking at either a linux2 or a linux3 binary. If file
tells you that the binary is statically linked, it will run on both
linux2 and linux3 systems; otherwise, you can tell which type it is by
running
ldd SOMEBINARY
If the string "libc.so.5" appears in the result, you're looking at a
linux2 binary, which will probably also run on most linux3 systems;
otherwise, you're looking at a linux3 binary, which won't run on
anything else.
> Also - I'd like opinions on the need for continued i386_linux1 support. I
> assume that right now the Linux infoagents users have mostly linux2 boxes,
> a small minority of linux3 boxes, and almost no linux1 boxes. Please let
> me know if you believe I have the proportions wrong, and whether we should
> continue to keep linux1 around.
That sounds about right, though I think you may be underestimating the
numbers for linux3. Abandoning linux1 should be safe, particularly
given that (almost?) all of the "linux1" binaries are linux2 or linux3
anyway.
--
Aaron M. Ucko, KB1CJC <amu@mit.edu> (finger amu@monk.mit.edu)
