Path: utzoo!censor!dybbuk!yunexus!ists!jarvis.csri.toronto.edu!mailrus!shadooby!samsung!uunet!pilchuck!dataio!bright
From: bright@Data-IO.COM (Walter Bright)
Newsgroups: comp.lang.c++
Subject: Re: Do class libraries have to be in source form (3 of 4)
Keywords: libraries, source, C++
Message-ID: <2219@dataio.Data-IO.COM>
Date: 22 Nov 89 19:01:30 GMT
References: <178@taumet.UUCP>
Reply-To: bright@dataio.Data-IO.COM (Walter Bright)
Organization: Data I/O Corporation; Redmond, WA
Lines: 73

In article <178@taumet.UUCP> mike@taumet.UUCP (Michael S. Ball) writes:
<Why do potential vendors of class libraries resist distributing source
<code?  Warning:  the following is filtered through my prejudices.

At Zortech, we distribute full source to the libraries that come with it.
The reasons are:
o	It really ain't that clever. Really, how complicated can strlen be?
o	Many users use the library source as examples of how to interface
	asm code to the compiler.
o	If there is a problem, the user can simply fix the code for his
	application, and continue, instead of waiting for an update.
o	Sometimes (though surprisingly rare), a user will find a bug
	in the library, and will propose a correct fix.
o	The extra money for the library source helps pay the bills.
o	We encourage users to modify the library if they have special
	needs (such as romable applications).
o	Library source helps third party library vendors be compatible
	with us.
The source to the graphics library is excluded from this because of licensing
problems.

<I heard the following reasons at "C++ At Work":
<1.  They will steal our source code.
	Who cares? A capable programmer can easilly disassemble your OBJ
	file anyway. There are several excellent programs that can do this.
	OBJ files are insufficient obfuscation to prevent it.

<2.  We can't maintain our code if the user can modify it.
	Programmers are well aware that if they modify it, they're on their
	own.

<3.  User's shouldn't want to see the source code.  They wouldn't
<    understand it anyway.
	Some of them want it so they *can* learn. Working, functional
	source code is great to teach people, as it is not a contrived
	example out of a book.

<I can speculate on a couple of other unstated reasons:
<4.  We'll loose our mystique.  Once people see how simple the code
<    is they will wonder why we charge so much.
	If it's so simple they would have done it themselves. Professional
	code just never turns out to be so simple. (I've not discovered
	a single customer who has figured out how my floating point code
	works! I don't even know anymore :-) and it's fully commented!).

<5.  Let someone else see how bad my code is?  Horrors!
	If you're that embarrassed by your code, maybe you're in the
	wrong profession!

<And a reason that they are so insistent on stating that libraries
<should be object only:
<6.  If we can persuade the competition not to distribute source
<    the users will have no choice.
	I decide for myself. If I see a competitive advantage to distributing
	source, I'll do it.

I can think of one pessimistic reason why some outfits won't distribute
source: the source was plagarized from someone else, and distributing
source would lay them wide open for a lawsuit. I wonder if this really
happens, and how prevalent it is.

I once had someone try to sell me source to a utility for my
compiler. Something about it looked suspicious (the comments didn't match
what the code was doing), and upon further investigation I discovered that
it was a disassembled version of one of my competitors' products!
Needless to say, I was very upset about that.

Another incident happened a few years ago when somebody tried to get Zortech
to OEM a MASM-compatible assembler. They wanted a lot of up front money.
The closer we looked at it, the more compatible it seemed. In fact, it was
so compatible that the only difference was in the copyright message!

The moral is, there are (unfortunately) crooks out there.