Remembering Paul Fillinich and Licensing C++ from AT&T | Hacker News Reader

Remembering Paul Fillinich and Licensing C++ from AT&T

24 pointsWalterBright4 comments

I was sad to hear that Paul Fillinich, an intellectual property lawyer for AT&T, passed away in 2020. I doubt many people are aware of his contribution to the success of C++.

Back in 1987 or so, C++ and Objective-C were neck-and-neck in the race to create a better C. I was interested in gaining a competitive edge for my C compiler (Zortech), and wondered which horse to get on. Stepstone owned Objective-C, and had some onerous licensing terms for it. But C++? I made some phone calls, and finally discovered that Paul Fillinich was in charge of IP at AT&T. I contacted him and asked:

1. can I get a license to create a C++ compiler?

2. can I call it C++?

Paul laughed. He said nobody had ever asked him a question like this in advance, they usually just went ahead and did things hoping nobody would notice. (Of course, being AT&T's IP lawyer, he did notice.)

Paul said sure, go ahead. You don't need a license from AT&T, and AT&T didn't trademark C++.

I owe him a large debt of gratitude, and so does the C++ community. Paul was one of the good guys.

This enabled a thriving ecosystem for C++, and we all know what happened to Objective-C.

I chose to implement C++ rather than Objective-C. At the time, about 90% of programming was done on MS-DOS. Bjarne's cfront implementation was not viable on MS-DOS because it did not support the near/far memory model required. After Zortech C++ (the first released native C++ compiler) appeared, C++ took off on the PC. I remember Borland's Eugene Wang was shocked at Zortech's sales, and Borland abandoned their plans for proprietary extensions to C and threw their lot in with C++, too.

I heard a bit later that Microsoft had also been internally developing their own object oriented extensions to C into a "C Star" language, but abandoned it after the success of Zortech and Borland's C++ compilers. I have never been able to verify this story, everyone I talked to said it was before their time at Microsoft.

Zortech did sell a lot of C++ compilers to Microsoft. It turns out that they were developing the COM technology with it. The ABI for COM exactly matched Zortech's virtual function dispatch mechanism :-)

I think, overall, you're right: the "open" nature of C++ (and C) has been a key factor in their success.

ObjC became available in GCC via NeXT (although somewhat despite NeXT) and so it was more-or-less equally available for many purposes. The language definition was always locked up though, and there was little innovation from Stepstone, NeXT, and later Apple. ObjC++, ARC, GCD, etc, were all useful but more-or-less proprietary.

Despite 15 years of being the native language for one of the world's most popular computing platforms, used by millions of developers, ObjC didn't leverage that advantage, and things like GNUstep have continued to have very little impact.

RIP Paul Fillinich.

Before Zortech C++ debuted, the volume of messages in the Usenet comp.lang.c++ and comp.lang.objective-c were about equal. Afterwards, the traffic in comp.lang.c++ took off, and the objective-c one stagnated.

Reply on news.ycombinator.com