Conference hydra::amiga_v1

2839.0. "I've had it with the Lattice crap" by KYOA::MIANO (When will Dallas get canned?) Thu Aug 17 1989 22:48

I have had it with the Lattice C Compiler.  It is the single biggest
piece of s&^t that I have ever seen from a packaged software product.
I am trying to convert a numerical program to run on the Amiga and the
compiler just can't do the job (and it's not a matter of size).

I seem to be up against the wall.  The Lattice compiler can't cut it.
For all I know the Manx compiler might not do it either.

I am ready to call Lattice up next week and take them up on their money
back offer.   Is there any way to do any arm twisting with them?
If I dump Lattice for Manx am I going to be back in the hole?

If have had the thing for about a month and I have called them up five
times already with bugs.  A small sample:

o Contrary to the product description it does not generate usable code
for the 68881.

o The debugger does not allow you to examine the floating registers.

o The debugger makes 3/4 of your stack unusable.

o The debugger often passes the wrong arguments to the main program.

o Some of the printf formating statements do not work.

o I have four separate c functions that crash the compiler.

o The assembler crashes when assembling the startup code provided by
lattice.

o The profiler does not generate correct output.

o The "-g" option does not always product a listing.

o pointer = (cast) malloc( sizeof( *pointer) ) ;
generates uninitialized variable warnings.

John

    
    Well...
    
    I know there are bugs in Lattice C. But they've been trying constantly
    to improve their product. For example, I received a patch disk two
    weeks after I complained about a certain bug with the (f)open()
    function. Manx also has bugs, but in contrary to Lattice, their
    last update is well over a year old.
    
    Now, what I'd do is to take your list of bugs and send them to the
    Lattice BBS. Or call their support line. Or send EMail to John Toebes
    or Doug Walker on UseNet.
    
    Make sure that you have the latest version. 5.04 should be out RSN.
    
    Talk to Lattice, I'm sure they'll listen.
    
    Regards,
    
    <CB>

Re: .0

> pointer = (cast) malloc( sizeof( *pointer) ) ;
>generates uninitialized variable warnings.

I'm amused by that bug.  That bug is indicative of a class of bugs
where the a C compiler fails to treat the expression being sizeof'ed
special.  No bad code is generated, but the compiler can issue
irrelevant messages because of it.  I think VAX C has a few bugs
of this type.

Anyway, a workaround is to use the LC option that allows you to 
suppress the uninitialized variable warning.  (You can suppress
any message from the Lattice compiler individually.)