mss+@ANDREW.CMU.EDU (Mark Steven Sherman) (01/21/88)
I see lots of people suggesting the use of GNU C++ as an implementation approach for toolkits. The last notice I saw about the GNU C++ distribution was something of the form "an alpha- release is available and use of GNU C++ is suggested only if you are willing to be faced with bugs and are familar with the internal structure of the GNU C compiler/backend". I don't know about anyone else, but we have enough arrows in our backs with initial (pre)releases of X.11 that I'd just assume wait a bit before committing any project to GNU C++, like rewriting Xtk or Andrew Tookit in GNU C++. -Mark
golds@rlgvax.UUCP (Rich Goldschmidt) (01/22/88)
Although it is free, GNU is not public domain. Their license contains some significant restrictions, which makes it unusable for a commercial product. While some might not consider this a problem, it will certainly prevent it from becoming a standard. This suggests it is not a good choice as a basis for Xtk. Rich Goldschmidt ARPA rlgvax!golds@uunet.uu.net UUCP uunet!rlgvax!golds UUCP (west) sun!sundc!rlgvax!golds
dcj%jacksun@Sun.COM (Donald Clark Jackson) (01/23/88)
Regarding the use/non-use of C++ for toolkits, I think there are more options for developers than I have seen discussed so far. 1) If you developing an object oriented toolkit today, and you wish to use C++, you could use one if the commercially available (& supported?) C++ translators, or you could use the 'alpha' version of GNU C++. 2) Regardless of the C++ translator used to develop the toolkit, users of the toolkit ought to be able to use any C++ translator that want. Developers should be able to switch translators as new (better, and/or cheaper) become available. Just because GNU C++ may not be stable today, doesn't mean that it won't be of production quality in the future. (This is a judgement each developer must make for him/her self) People developing object oriented toolkits have a decision to make 1) Does C++ have the features I want/need, versus the cost in time, effort, support, and non-standardness to re-invent my own object system. * I can start using C++ today using a commercial C++ translator, and bet that GNU C++ will be a viable option in the future. * I think toolkit developers should think hard before commiting themselves to re-inventing an object system just because GNU C++ isn't stable right now. You could start with a 'stable' commercial C++ translator, and switch to GNU C++ later. (The only problem is if you can't afford a commercial C++ translator to get started) * Availability of GNU C++ may not be an issue for developers of commercial products.
wesommer@athena.mit.edu (William E. Sommerfeld) (01/23/88)
In article <802@rlgvax.UUCP> golds@rlgvax.UUCP (Rich Goldschmidt) writes: >Although it is free, GNU is not public domain. Their license contains some >significant restrictions, which makes it unusable for a commercial product. >While some might not consider this a problem, it will certainly prevent it >from becoming a standard. This suggests it is not a good choice as a basis >for Xtk. Wrong. The license for GNU CC (and GNU C++) states only that you are compelled to distribute source to the compiler, or an offer to distribute source to the compiler in exchange for a distribution fee; it does not say that anything compiled with it must be covered by those terms, and GCC doesn't put a GNU `copyleft' into its output. Also, the bizarre licensing `restrictions' on GNU C++ are irrelevant to whether or not an object oriented user interface toolkit should be written in C++; the existance of GNU C++ merely means that it is now a lot easier to get a hold of a C++ translator for some machines. - Bill
golds@rlgvax.UUCP (Rich Goldschmidt) (01/25/88)
In article <2479@bloom-beacon.MIT.EDU>, wesommer@athena.mit.edu (William E. Sommerfeld) writes: > In article <802@rlgvax.UUCP> golds@rlgvax.UUCP (Rich Goldschmidt) writes: > >Although it is free, GNU is not public domain. Their license contains some > >significant restrictions, which makes it unusable for a commercial product. > > Wrong. The license for GNU CC (and GNU C++) states only that you are > compelled to distribute source to the compiler, or an offer to > distribute source to the compiler in exchange for a distribution fee; > it does not say that anything compiled with it must be covered by > those terms, and GCC doesn't put a GNU `copyleft' into its output. > > Also, the bizarre licensing `restrictions' on GNU C++ are irrelevant > to whether or not an object oriented user interface toolkit should be > written in C++; the existance of GNU C++ merely means that it is now a > lot easier to get a hold of a C++ translator for some machines. > > - Bill My reading of the license restrictions accompanying GNU are that any derivitive products must be distributed free, with source code. I am not a legal wizard, but I would not be willing to risk a large software development on the interpretation of the terms of the GNU license. If the FREE SOFTWARE FOUNDATION wishes to clarify the intent of their restrictions as a part of their license, that might shed some light on this issue. I do think that the combination of a commercially available C++ and GNU C++ makes C++ a reasonable possibility as a target. However, I would hope that toolkits would not be built to depend on an unsupported offering, if possible. It is also true that developers of both Xtk and Andrew toolkits have already posted why they are not too excited about C++. Rich Goldschmidt uunet!rlgvax!golds sun!sundc!rlgvax!golds rlgvax!golds@uunet.uu.net
yba@athena.mit.edu (Mark H Levine) (01/26/88)
I do not believe that either a reasonable layman nor a legal expert would assert that a program compiled under, say GNU C++, is in any way a derivative work based upon the translator. This seems to be the basis of your difference with Sommerfeld, and can probably be safely dismissed. (If you did a commercial compiler by starting from the base of GNU's body of source code, you are in the other camp). Can this now be moved to private e-mail or directly to RMS?
bilbo.geoff@SEAS.UCLA.EDU (Geoff Kuenning) (01/26/88)
In article <2479@bloom-beacon.MIT.EDU> William E. Sommerfield writes: > >Although it is free, GNU is not public domain. Their license contains some > >significant restrictions, which makes it unusable for a commercial product. > >While some might not consider this a problem, it will certainly prevent it > >from becoming a standard. This suggests it is not a good choice as a basis > >for Xtk. > Wrong. The license for GNU CC (and GNU C++) states only that you are > compelled to distribute source to the compiler, or an offer to > distribute source to the compiler in exchange for a distribution fee; > it does not say that anything compiled with it must be covered by > those terms, and GCC doesn't put a GNU `copyleft' into its output. Wrong yourself. Such a deal, if I want to use Gnu C++ I am forced to be ready at any time to make a large and complex source distribution available to my customers; among other things this opens me up to the possibility of legal action from them should Gnu C++ fail on them. One only has to read the Gnu Manifesto (or Stallman's more recent allegation that the world's security needs are identical to those of MIT Lincoln Labs) to realize that the Free Software Foundation has some staff members who are, shall we say, slightly separated from the realities of the larger world. One of those is the reality of needing more flexible licensing. > Also, the bizarre licensing `restrictions' on GNU C++ are irrelevant Note that "bizarre" is Bill's word, not mine. (This doesn't invalidate his point about the irrelevancy, however). Geoff Kuenning geoff@lcc.ucla.edu geoff@ITcorp.com