how brain dead is AIX?

[Steve DeJarnett]

In article <508 at ss01.pppl.gov@ccc.nmfecc.gov> jsm at ss01.pppl.gov@ccc.nmfecc.gov (John Scott McCauley Jr.) writes:
>I have a copy of 'IBM RISC System/6000 Software Offerings Overview' brochure.
>I am a little worried about two of the statements made: 
>
>A) When there are conflicts, they are resolved with priority 1) POSIX
>	2) SVID 3) 4.3 BSD 4) AIX/RT. (p. 3). 
>
>   I gather that AIX is System V kernel with the Berkeley stuff emulated.
>   So, do you get:
>     1) the System V file system or the 'Berkeley Fast File System'

	Neither.  You get the new Journaled File System (JFS).  Basically it's
designed to ensure that you don't lose any files when the system crashes (all
file creation is logged on a special logging filesystem).  Thus, if the system
crashes, upon reboot it simply goes through all of the logs and makes sure it
has all of the files.

>     2) streams or sockets

	Sockets.

>     3) job control (control-Z)

	You bet.

>     4) AT&T or Berkeley terminal device driver.

	Actually, what you get is a POSIX line discipline and a BSD line
discipline, although almost the entire BSD line discipline was duplicated in
the POSIX (default) line discipline, so unless you're using a Hazeltine 
terminal with no '~', you'll probably never notice.  Also, be aware that in
4.4 (or whatever it will be called), CSRG has said that they will be shipping
a POSIX-compliant line discipline, and will be deprecating the BSD line
discipline.  I believe Sun is already shipping a POSIX line discipline in SunOS.

>B) 'Most 4.3 BSD commands, system calls, and library routines are supported.'
> 	(p.3) and features include a '4.3 BSD compatibility library'.  (p.5)
>
>  Ok, what is missing? 

	Briefly:

Libraries: getmntent, addmntent, etc. (the *mntent functions), endusershell,
	getdiskbyname, getusershell, ns_addr, ns_ntoa, quota, setquota,
	setusershell, tell, x3wind.

Commands:  arff, bad144, badsect, dcheck, diskpart, drtest, error, edquota,
	gcore, kgmon, lpc, mset, quot, quotacheck, quotaoff, quotaon, 
	repquota, systat, trsp, and XNSrouted.

	Fairly simple workarounds are supplied for all of the *mntent routines.

>One more question: we already have several BSDish systems (Dec- and Sparc-
>  stations).  Things are currently seamless: all programs we have needed so far
>  will compile on both and run fine on both. Would the same be true of an AIX
>  system? Will we have to hire someone full time to support the AIX? How many
>  lines like '#ifdef AIX' are there in the source code for gnu-emacs? X11R4?

	Well, there are 2 fairly comprehensive documents shipped with the 
system just for those of us with BSD backgrounds.  There is a document telling
you what to expect from the header files, how to set up the compiler to behave
the way you want, and describing the differences between AIX Version 3 and BSD.
There is another document that describes the differences in administering an
AIX Version 3 system versus a BSD system, with detailed discussions on:

	Accounting, Backup, Boot and Startup, Cron, Devices, Finding 
Information, Hard Disks, Paging Space, the Network, Passwords, Performance
Measurement and Tuning, Printers, the System Resource Controller, Terminals,
and UUCP.

	The Free Software Foundation has indicated that they will ship the 
necessary config files for Emacs, so you should be able to build it the same
way you would for a DECStation, Sun, or any other FSF-supported system.  The
current AIXWindows product is X11R3-based, with several extensions.  Since it's
a product, we haven't given it away to the Consortium.  :-)

>		Scott

Steve DeJarnett			Internet: steve%ibmsupt at uunet.uu.net
IBM AWD Palo Alto		UUCP:	  ibmsupt!steve at uunet.uu.net
(415) 855-3510	  These opinions are my own.  I doubt IBM wants them.......