Chewy@cup.portal.com (Paul Frederick Snively) (08/09/90)
Hi, y'all! I'm sure this is a rather stupid question, but I need to know... To what extent are AT&T CFront 2.0 and g++ source-code compatible? The question, of course, is whether a forthcoming project can be done in g++ using existing documentation, such as the Lippman (sp?) text, without running into headaches. Issues such as whether g++ and AT&T CFront construct virtual tables the same way, etc. are irrelevant. Any insights are appreciated. Thanks! Chewy@cup.portal.com
keith@csli.Stanford.EDU (Keith Nishihara) (08/09/90)
In <32603@cup.portal.com> Chewy@cup.portal.com (Paul Frederick Snively) writes: >Hi, y'all! >I'm sure this is a rather stupid question, but I need to know... >To what extent are AT&T CFront 2.0 and g++ source-code compatible? It is generally a good idea to look before you leap. I can't believe you site expires articles so fast that the following four were not available when you posted your question: a Jim D 5 11-Y-03 14 porting tips for c++ b Jim ADCOCK 10 > c George Wu 118 > d Keith Nishihara 79 >> The article by George Wu and my followup (as Keith Nishihara) address your question. In summary, g++ and cfront 2.0 are so similar that with reason- able care you should have no problem. I recently ported a 35000 line application written under g++ on a DecStation to cfront on a Sun. Total time was 10 hours, and most of that was recoding to fix an incompatibility in the event ordering between the X window implementations on the two machines. My advice is to try the port when you have about 5 to 10 k lines of code, learn the in- compatibilities, and then you will have no problem. My impres- sion is that g++ is a more faithful implementation of the implied standard (with extensions discussed in the gnu documentation), however, that is probably because I hit all the problems in the cfront implementation at once. Neil/. Neil%teleos.com@ai.sri.com
bashford@scripps.edu (Don Bashford) (08/10/90)
In article <14826@csli.Stanford.EDU> Neil Hunt <Neil%teleos.com@ai.sri.com> writes: >In <32603@cup.portal.com> Chewy@cup.portal.com (Paul Frederick Snively) writes: > >>Hi, y'all! >>I'm sure this is a rather stupid question, but I need to know... >>To what extent are AT&T CFront 2.0 and g++ source-code compatible? > >It is generally a good idea to look before you leap. I can't believe >you site expires articles so fast that the following four were not >available when you posted your question: > a Jim D 5 11-Y-03 14 porting tips for c++ > b Jim ADCOCK 10 > > c George Wu 118 > > d Keith Nishihara 79 >> I don't know about him, but on *our* machine they are expired. Could someone e-mail them to me? I'm also interested in the question of libg++ portability. I recently tried to give my code (including the necessary libg++-include files and genclass generated code) away to someone who has only a cfront clone and it could not deal with libg++'s use of GNU extensions such as the use of named return values in String.h. So my questions are: 1) Is there a cfront-compatable version of libg++ available, or better yet, a program to convert g++ extended code into standard c++ (alright maybe I'm dreaming). 2) How hard is it really to get libg++ stuff to compile under cfront I'm not in a position to explore this on my own, since g++ is the only c++ I've got (and I think its great!). Donald Bashford bashford@scripps.edu