🚀 go-pugleaf

To: 
Path: lsuc!jimomura
From: jimomura@lsuc.UUCP (Jim Omura)
Newsgroups: comp.sys.m68k.pc
Subject: Re: (none)
Summary: Defining "good"
Message-ID: <1826@lsuc.UUCP>
Date: 28 May 87 00:41:34 GMT
References: <PEPRBV.870526.122938.A0@CFAAMP.BITNET>
Reply-To: jimomura@lsuc.UUCP (Jim Omura)
Distribution: world
Organization: Consultant, Toronto
Lines: 106


In article <PEPRBV.870526.122938.A0@CFAAMP.BITNET> PEPRBV@CFAAMP.BITNET (Bob Babcock) writes:
>jimomura@lsuc.UUCP (Jim Omura) writes (about OS-9):
>>     There is a good selection of development
>>tools, as well as application software.
>
>I find this hard to believe.   It took 2.5 years to get a Fortran
>compiler for our OS-9/68K system.  As far as I know, Microware is
>the only supplier of compilers, so if they don't have it, you are

     Even as I wrote the original comment I knew we'd have
to get more specific about it, but I was tired.  Anyway, what's
"good" selection?  Notice I didn't say "staggeringly abundant"
or "beyond your wildest dreams" or anything nonsensical like
that, but I stand by my statement of "good".  What I mean by
"good" is that there is a sufficient selection of working tools
for what most programmers find themselves up against.  I also
weighted the fact that most of the Usenet population (and those
on other Nets where these messages might end up) tend to be
C oriented.

     OK.  Bottom line for most of "us":  The Microware C compiler is
"good".  I'm not a super C hacker, but I've used it enough to confirm
that it works and covers the majority of C programmers requirements
which is evident when you start porting stuff.  I've done a couple
of ports and looked over a quite a few.  It's not perfect and I have
my own gripes, but we are shooting at a moving target and things are
constantly improving with it.  The current version is 2.1 and the
code is, depending on the compiler options, fairly fast as is evident
by benchmarks on my '020 machine, or extremely compact, due to the
efficiency of the system, or some compromise of the two.  For more
detail, the original inquirer can get a copy of the manual (which I'm
assuming you have ... ).

     As for other languages, Omega's Pascal has been out for at least a
year as far as I know and is supposed to be quite good (I've never used
it), Microware has a Pascal and also a BASIC, which is currently being
redone a bit.  There is at least one other Fortran coming out fairly
shortly (if it isn't available right now) which should be better than
the Microware Fortran if what I hear is correct.  I won't argue that
this is "enough" because everyone has their own standard for "enough".
Heck, the more the better.  As for me, I don't see myself using more
than one or two languages in the future, so you know where I stand on
"enough".

>stuck.   Also, most of the little  utilities  which are available
>for MS-DOS don't seem exist for OS-9.  I would be much happier if

     Utilities are strange to compare.  If you don't have one good debugger,
you make do with 6.  This truth is hard to believe until you move from
a powerful tool to a less powerful tool, or a powerful system to a less
powerful system.  I can't compare systems.  Let's look at a typical problem:
You've been doing a series of compiles and have a bunch of versions of
the same program (object code) in a directory.  Under some systems, you'd
need a 'cmp' comparison utility and maybe a checksum utility to figure
out which were in good shape, and hopefully you set your date right
every time you booted.  Well, OS-9 has a comparison utility and the CRC
is embedded in the file.  The date is also there (again subject to your
diligence :-).  It's got that kind of stuff pretty much out of the box.
It is possible to treat a device as a file pretty much with '@' addressing
mode.  There's work being done on a screen oriented debugger instead of
the current line oriented debugger.  There is a good array of file
arranging and manipulating stuff (mass copy, grep, directory tree reports,
etc.).  Furthermore, we can tap most of the C sources written for Unix
due to the similarity of user interface.

     History:  The standard shell doesn't have history.  I dunno.  Again
I'm not running around screaming for history.  Then again, I'm a special
case.  I use an Atari 1040ST for a terminal.  To me, the 'history' that
most people think is so wonderful is a pathetic joke.  My "history" is
about 800K with screen editing courtesy of the Flash! telecom program.
If you want to write a new shell with the usual 10 - 20 lines of history
go ahead.  As for me, I tend to keep a 1/2 day's work available for
review at any given time.  Someday, I expect more people to work with
systems closer to mine.  The price of the Atari 1040ST and Amiga 500 are
around what VT-100's costed at one time.  You just gotta try it man!
Beyond that, well, it boils down to "you show me the problem and we'll
see if I got a solution".  No real way to compare.

>I could have equivalents for just the Norton Utilities and CED (a
>command line editing/history utility).

     Lastly, I also considered "applications" level tools.  We have
two database/application systems which I hear are pretty good.  CSG IMS
and Sculptor.  I'll be buying Sculptor shortly, but I have contact
with people who have been using it and feel confident that it'll cover
the needs of most people doing business databases.

     Again, there are two problems with this discussion.  First, "good"
is whatever you want to say is "good".  In quantity and quality, I think
say, 80% of us will find pretty much everything we need either supplied
with version 2.0 OS-9 68K (Professional) or reasonably easily obtained,
subject to the fact that most people on the Net are programmers in the
first place and so we should be able to do a bit of the work to put together
*some* of the tools *some* of the time.  Secondly, we're talking about
another "moving target".  Did you know that the current Shell has a PATH
variable?  If the last version of OS-9 68K you saw was 1.2 then you
might not know that.  Nor would you know about some other changes
such as .login and .logout files, 'group' level permission support
and a number of other changes.  Say in about 6 months, for all I know,
we could have 'history'.  It wouldn't take much to add it.

Cheers! -- Jim O.
-- 
Jim Omura, 2A King George's Drive, Toronto, (416) 652-3880
ihnp4!utzoo!lsuc!jimomura
Byte Information eXchange: jimomura