AIX not UNIX

[Paul Pomes - UofIllinois CSO]

So what's wrong with our new IBM RISC-6000/540 box?  Besides the AIX OS
installed on it, nothing.  AIX represents "Not Invented Here" carried to
an extreme.

Case 1: The question arises: "What debuggers are available?".  First try
"man -k debug".  Response is "/usr/man/whatis: File not found".  No problem,
run /usr/lib/makewhatis.  It runs very quickly and produces a zero length file.
Reason: AIX has no man pages!  Instead it has this huge _encrypted_ database.
No more grep'ing the man pages for clues.  According to the man executable
I committed a criminal offense finding this out:

 All Images, Data Structures, Algorithms, and Machine Code Copyright (c)
 1985 Activenture Corporation, Monterey CA. 1985 All Rights Reserved. Any
 investigation or use of this information without Activenture written
 permission is considered criminal action.

Reality fault: core dump  --  Products with a built-in attitude problem I
can live without.

No man pages is a serious defect.  I no longer wonder why it's called AIX
instead of UNIX - it's not UNIX, it's "UNIX-like".  UNIX-like is license
to delete standard parts (man pages, fortran, etc) that you then sell as
"extras".  Bleeech.

Case 2: If DELete is your erase character, don't expect it to work in vi like
it does on other systems.  It ALWAYS deletes the previous character regardless
of whether you're in insert mode or not.

Case 3: NOTHING so far has compiled and ran.  I have programs that work
unchanged on 4 different systems yet require special #ifdefs to work on
AIX.  I can't even find a system specific #define in either the compiler
or system /usr/include files.  That requires editting the Makefile to add
my own -DSUB_STANDARD

Case 4: Many well-known values have been senselessly re-named.  There's no
longer a value of BSD in the /usr/include files, it's now _BSD.  edata and
etext have also been re-named to _edata and _etext.  How nice.

Case 5: No support for re-sizing windows.  Other vendors have had this for
a year or more.

Case 6: Several typedefs and #defines have been gratuitously relocated to
different include files.  fd_set, previously in sys/types.h, now lives in
sys/select.h .

Case 7: Where's the malloc.o object?  It's not in any library I can find.
The evidence points for it to be somewhere in the kernel.  Wonderful.  I
suspect that it's trashing my data segment but I can't debug it.

case 8: No source code.  It isn't UNIX w.o. source code.  IBM is now telling
us next June for our source tape.  That's 8 months away folks, how about
this November instead?

/pbp
--
         Paul Pomes

I'm official (officially myself).  Accept no substitutes unless they pay cash.

UUCP: {att,iuvax,uunet}!uiucuxc!paul   Internet, BITNET: paul at uxc.cso.uiuc.edu
US Mail:  UofIllinois, CSO, 1304 W Springfield Ave, Urbana, IL  61801-2910