CompuServe Thread

Larry, I'd have to say that there was never a time when "most developers were using manx". There was a time when manx was highly touted here on CIS, despite it's backwards inclusion of garbage capabilities like ILAC, it's inability to link in the std Amiga format, the all-in-one (small) data capability, it's far-from compliance with ANSI, and it's default to 16 bit operation when that was NOT the std… I heard a lot of people here singing it's praises – but I rarely ran into code that was actually made with it. I know some of the PD guys use it – Keith, for instance – and I know that some of the developers use it, too. There were some perfectly horrible code examples in Transactor, for instance. But Lattice was never the minority compiler – it only seemed that way to those who had Manx, because they wanted it to be that way. they generated smaller code only with the default options set to non-std – 16 bit. Lattice pointed out (repeatedly) that if you asked for 16 bit stuff and so on with the then-current compiler, you got executable of comparable size. Part of that issue was start-up code, too – the Lattice startup code is much more robust, feature wise, than the Manx, which made it bigger. Faster code? Same issues. Use the same options, you got comparable code. There never was any huge performance gap, except to hopeful Manx owners. I've been generating Amiga applications with Lattice for almost five years now, and I've been completely satisified – my one fling with Manx was truncated by the peculiar proprietary linking, the lousy defaults, the small data capacity, and the fact that it supported ILAC, which I view as the worst C sin going. Ben Amateur Radio Callsign is A A 7 A S

You're free to disagree – but you're wrong. :^) Ben Amateur Radio Callsign is A A 7 A S

No, manx never had the better compiler. Manx defaulted to 16 bit generation (non-std at the time) which resulted in considerably smaller executeble. Naturally. They also had a smaller, less capable startup module that was linked in – again, we end up with smaller code. They had a proprietary link format. You couldn't link in an AMiga .obj file. They were WAY behind on the current "in-vouge" C stds, that is, the stds that were in the proces of becoming ANSII, which Lattice kept up with very well. There was a lot of BS here about how Manx made this wonderful small, fast code. John Tobbes, who wrote the Lattice compiler (more or less) took the time to show the Manx users that using the compiler switches to select 16 bit, no stack checking, and so on, that the Lattice compiler benched out the same size or SMALLER!!! Further, he provided a number of examples where the startup code was stripped to the functionality of the Manx code, which even reduced the cmd size more. The alleged superiority of Manx is mainly propaganda. Things like defaulting to base register addressing… sure, saves code size all over the place, AND limits you to 64k (or sometimes 32k) of data. Lattice defaulted to 32 bit addressing – more reasonable, IMHO, but you could always tell the compiler you wanted base-relative addressing and get it, along with the penalty of small data space. Lattice got tired of hearing all this trash, and made the base relative, 16 bit options the default. Too bad – it was not a compiler change, just a default change, and that only because of all the Manx propaganda from all directions. The manx compiler follows a philosophy similar to the IBM one… we'll do it our way, never mind the standards. I have little use for it. Lastly, Manx supports in-line asm code; there is nothing worse in the world of C programming that in-line asm. It's a poor technique used by lazy or incompetant programmers (or both). Ben Amateur Radio Callsign is A A 7 A S

I don't think anybody can complain about the competion between Manx and Lattice! I think we're all winners (and whiners).

Mike, I started out using ,the Amiga-C (Lattice) compiler. I finally gave up on doing anything serious with it and went to Aztec-C. I love it. I had already sunk the money into the Amiga-C compiler, and I wanted to stay with it (you know, the baby duckling syndrome), but Lattice didn't make it very easy on me. I called for support a couple of times, because it appeared the compiler was generating incorrect array and structure offsets, but was told I would have to get support from Commodore because I had purchased the Amiga-C compiler, not the Lattice compiler. This was a big mistake on Lattice's part, because it alienated programmer's who had purchased the compiler, knowing it was developed by Lattice, and expecting Lattice support. I decided to buy the Lattice product for the support, but I read an ad that announced the introduction of the Aztec-C source level debugger. That was a real coup, fella's. When I called Lattice and asked about a possible SDB offering from them (I was STILL trying to buy Lattice!), they told me that NO enhancements were planned for the compiler. Period. What was I to do? I bought the Aztec compiler and SDB, and I have been happy ever since. I have worked with Unix and I am fond of vi, grep, etc. This was icing on the cake. To be honest, the 5.0x release of Lattice is a far different/better animal than the old Lattice compiler I was working with, but the Aztec C DEFINITELY was superior in a number of ways before Lattice got back into the fray. BTW, I have used a number of different Aztec C cross compilers, on the job (8080, Z80, etc.), and they have been around a LONG time (stability, etc.). Lattice has been around a long time too for that matter. So either compiler will do an EXCELLENT job, its just that the Lattice marketing guys burned me. BTW, when I was at AmiExpo 2 years ago, a couple of their programmers were showing me their soon-to-be-released SDB, and had me really interested. Then along came the sales force, who got bent out of shape, and made them stop talking to me! I'm sorry Lattice, but I'm [ MORE ]

[ continuation ] not going to BEG you to let me buy/investigate you products. 'nuf said 😉 BTW, I just got my Aztec-C 5.0 upgrade. WOW! Marcus