Path: utzoo!attcan!uunet!zephyr.ens.tek.com!uw-beaver!mit-eddie!bloom-beacon!eru!luth!sunic!tut!funic!funic!vinsci From: vinsci@soft.fi (Leonard Norrgard) Newsgroups: comp.sys.amiga.tech Subject: Re: How about a commercial GCC for the Amiga? Message-ID: Date: 26 Jun 90 01:13:23 GMT References: Sender: vinsci@funic.funet.fi (Leonard Norrgard) Organization: Soft Service, Inc. Lines: 78 In-Reply-To: dillon@overload.UUCP's message of 19 Jun 90 23:11:23 GMT > It makes little sense to port GCC to the Amiga. First of all, the >main compiler pass itself is only a small part of a C compiler >distribution... It is doubtful that porting GAS (the GNU assembler) would >be much easier than writing one yourself and you still have to deal with >the linker, C library, and includes, as well as other items. The possibility >of easily porting GDB is almost nil (that is a guess, not having tried to >do it myself). Hey! Not so fast, please! First off, there are other reasons to port GCC to the Amiga, other than to get yet-another-C-compiler. Like a C++ *compiler* and zillions of programs written using GCC. Lattice C, for example, can't handle most larger programs without lots of hacking to get it to accept the source. I really hate random limits on #define sizes etc. Also, it would be interesting to compare the Lattice optimizer with the GCC one. I'm working on a GAS port rather than writing a new assembler since GCC & GAS are already related and will be even more in the future (when the intermediate text interface dissappears). Also, having gas to make AmigaDOS object modules means that one can cross-compile on a faster system simply by running gcc&gas on the faster system. Sure, there's a lot to a C system. But we've got to start somewhere, haven't we? With GCC & GAS running, you've already come along way. Add a current commercial compiler and you have libraries, include files and a linker. Of course, free libraries, includes and a linker will be written sooner or later too when someone actually gets around to it. (C-A: how about making ALINK source available? Or is anyone actually using it still? ;-) It would be put to really good use!) > Secondly, you cannot port GCC direct to the Amiga... I tried this and >the executable is on the order of 600K which is unusable. What's wrong with this? Sure, it's not small, and sure, it's not for everyone. It still gets the job done. > Not only that, >but most GNU stuff takes a lot of memory run-time as well. True, but as pointed out, 2.5Meg is enough to run GCC. Those who want GCC will get the extra memory. Those who are not prepared to get more memory will have to live with whatever compiler they have, at least noone can complain on GCCs price! > It would take >a lot of work to port GCC and the source itself is so complex that even if >you do manage to get something running I can only wish you good luck in >tracking down and fixing bugs. Of course it takes a lot of work, that's why we formed a working group to accomplish the port. And Matt, you're welcome any day! The only bugs we will *have* to find ourselves are those not present in the original GNU code, for those, a bug report to prep is sufficient. The GCC porting project can be reached at: (so far about 25 people) amiga-gcc-port-request@nic.funet.fi Let's see, that should be ...!uunet!nic.funet.fi!amiga-gcc-port-request if you're using UUCP addressing. That is the place to ask to get added to the mailing list etc. [Now, to those who might want to join in on this project, please consider before contacting us: a) have you got the time b) have you got the skill ] There's a mailing list set up on nic.funet.fi for the group and FUNET (The Finnish University & Research Network) have donated "enough" FTP:able disk space, so distribution is not a problem. An Amiga GCC has been reported to run on the Amiga, *and* to generate correct code. See? It's not impossible, it just takes a long time... Now, there's still lots of work to do before it's ready for a general release, so don't hold your breath. > -Matt -- Leonard Norrgard GCC port coordinator