CompuServe Thread

#C Compilers: which one?

4 messages in this thread
#58447From: Black Belt SystemsJul 23, 1989 2:50 PM
The heck we would. <grin> I like the fact that Manx is there – for the competativve reason you mentioned. Your points are well stated, and I accept them, sort of. I will say that I am very glad I don't have to switch compilers, simply because the environments are quite different. Should the Manx compiler become the superior compiler (which would surprise me – it's never been that yet, despite the reviewers hype to the contrary) It would have to offer something spectacular to make me uproot all the megabytes of source I have. Pascal? Ech! Perhaps Modula <turning green> –Ben-
#58524From: David HamkinsJul 23, 1989 9:46 PM
Ben, When Lattice first came out it was a dog. Manx was far superior to Lattice. To claim that Lattice has always been number one is just plain naive. If you doubt me, whip out a copy of the original Lattice compiler and compile the following: main() { printf( "Hello out there!\n" ) ; } After about 3 minutes, you should get an object module. Manx's first release was an order of magnitude faster (and better). It's just recently that Lattice has caught and surpassed Manx. Dave @OTG
#58552From: Black Belt SystemsJul 23, 1989 11:52 PM
That entirely depended on how and which way you linked it, which print modules you brought in, and so on. I used the original (v1.0) Lattice compiler to create "PCLO", and was quite satisfied. A printf is no indication of the compiler – a complex app is, however. At the time those benchies were making the rounds, the entire prototype of PCLO was smaller than that default generation. What you were getting was a big chunk of library, which is no indication of the compiler's efficiency at all. In the beginning, btw, there was no Manx compiler – which sorta made Lattice a better choice, neh? <grin> Also, even given the 16 bit defaults that manx had (while lattice defaulted to the bigger, more powerful (and slower) model, and the fact that you couldn't (at the time you refer to) link the manx output with regular Amiga object modules, the Manx compiler was all but useless for wide-scope development. You had to use their environment, which was nothing but an incompatable port at the time. Again, that made Lattice better. You have to be very careful when you quote a specific benchie to say compiler a is better than compiler b. All that cr*p from the Manx camp about the problems with Lattice seemed to have no effect on the programs I produced with the compiler. I have run into very few bugs, and I've been using the compiler(s) from Lattice since the day the machine was released. We (SoftCircuits) showed the very first CAD program ever released on the Amiga at COMDEX within 7 months of the machines release, and we had it selling at the time. AEGIS was there, but had many bugs and couldn't sell anything. Since then, many Lattice-compiled products have crossed my threshold on the way out, and seldom have I had cause to complain. As a result, I view a great deal of the bug-crying that Lattice endured as Manx and Manx-user inspired BS. I have my experience to draw on – I've been extremely sucessful in using the compiler in many ways. –Ben–
#58587From: David HamkinsJul 24, 1989 6:21 AM
Ben, The fact of the matter is this. I have yet, in all my work, run into exactly one bug with Manx. With Lattice I've found half a dozen. If I *wanted* to find more bugs in Lattice, I could. Just pick any feature that was recently introduced and start running stuff at it. However, Lattice is stable in the core. Stick to the standard stuff and you seem to be ok. For this reason I feel comfortable recommending Lattice over Manx. Nevertheless, I'm anoyed by all the bugs. I cannot for the life of me believe that you honestly think the original release of Lattice was better than the original release of Manx. With Lattice you could watch the paint dry, the grass grow, and the Manx setup compile and link 2 maybe even 3 or 4 times while its counterpart staggered on. Dave @OTG