21 Apr
2002
21 Apr
'02
5:06 p.m.
It was thus said that the Great Richard Erlacher once stated:
That's not been my experience. In fact, until LINUX, which I don't presently
use either, I had never seen any software of any use at all other than for
software development for UNIX. The EDA stuff I saw was not terribly useful,
but some folks managed to beat it into submission. I can't forget the
tradeshow when I ruled out UNIX in my mind. A vendor had essentially the same
software for DOS and UNIX. The UNIX version cost 50x what the DOS version
cost, and the hardware also cost over 10x the cost of an adequate PC. The two
software packages "looked and felt" as well as worked, indistinguishably once
one was inside the application.
I did some work for a company who's only product was originally written
under MS-DOS, and I was hired to help with a UNIX port of the product. They
were planning on selling the UNIX version for about 10 times the MS-DOS
version even when (after the port) the source code is the same (except for a
few portions that hit hardware and was fairly isolated). They even refused
to sell a Linux version (even though that's the system they used for the
initial port) because they were afraid of not being able to sell it at 10
times the cost.
The reason for the disparity? Because UNIX customers are accustomed (or
were, maybe still are to a degree) to paying such prices and would view a
product that was too cheap as not being good enough. And they wanted to
offset the higher support costs that might come with the UNIX version. They
refused a Linux version for as long as they could, because they knew they
could not sell their product at their inflated price for Linux (as most of
their customer for the Linux version would be customers from their MS-DOS
version moving over to Linux).
Another company I did work for (helping with the UNIX port of their
primarily Windows product) sold the same product, from Windows at about
$1,000 a copy, to a Stratus mainframe for $50,000 a copy. Same product.
Same code base (for the most part). Why? Because they could. And the
$50,000 version fit on a single floppy.
and, for the most part, the freeware is often better
than the commercial
products. I've seen little "source-available" freeware that was very
good,
however. The LINUX stuff is a good example. Much of the code sits, full of
ugly hacks and undocumented modifications, among comments relevant only to the
original code that was abandoned six or seven revisions back, though it's not
obvious. It's a wonder any of it works, but it seems it does. It's unlikely
there'll ever be UNIX/GNU freeware that's as useable as the comparable
DOS/Windows stuff, though, since what looks to be the case is that nobody
wants to document the UNIX/GNU freeware.
So you are assuming that commercial code is somehow better than UNIX/GNU
code? Having worked at a few commercial software companies (except for IBM,
always hired to port code from MS-DOS/Windows to Unix) I can tell you that
the commercial code is full of ugly hacks and undocumented modifications,
among comments (when they actually exist---the first company I mentioned
above had, as part of their coding standards, mandated that no comments be
added to the code) relevant only to the original code. At the second
company I mentioned, the core of the product was written and maintained by
*one guy* and only *one guy* because no one else in the company could
understand the code. I saw the code in question and yes, I could see why no
one else wanted to touch this code. So code quality is just as bad (or
worse) in commercial products.
-spc (Was horrified at how bad commercial code is ... )