Conference hydra::amiga_v1

The preceding information was downloaded last night from the Lattice BBS.
Another note said that they will be sending update notices to all registered
users. Also, you may order your upgrade by phone using a credit card.

At last, a source level debugger......yeah!!!

Frank

    
    I received my Lattice V5.0 update two nights ago.  I haven't had
    much time to really dig into it but here are my first impressions:
    
    1.  Excellent documentation!  Two of the small 3-ring binders. 
        The presentation looks very professional.  The content looks
    	good too, but like I said I haven't really dug in yet...
    
    2.  CodePRobe (aka CPR) worked great on the small programs I tried.
        To use VAX Debug terms you can:  Go, Examine, Deposit, Set Break,
        SET BREAK where WHEN (exp) DO (exp).  You can also break when
        a particular variable changes.  Multiple break points are allowed.
    	You can examine variables defined as structures and it will
        show the structure contents.  You can also get type definitions.

    3.  Many other convenient features including:
    
          - C, ASM, and Mixed display.
          - Breakpoints can be set with the mouse.  "Go" can be performed
    	    with a double click on a source line.
    	  - Compiler companion included.  Lattice Screen Editor is
            included and is coupled with the compiler.  You can compile
    	    from within LSE and it will goto the errors in the source.
            (I didn't have a chance to try this yet.)
          - They've defined exactly what needs to be done to make a
    	    program reentrant so that it can be used with "resident".
            In fact most of the tools including the compiler, editor
    	    and linker are all "pure".
    
    I really could go on and on.  Are their any features in particular
    someone would like me to check out?
    
    regards,
    
    steve

    I just got home and found mine waiting at my door.  Now to see how
    much better it is than V4.

Has anyone tried running the first example given in the Debugger manual
(lines.c)?  I compiled/linked it with: "lc -d3 -La lines" then ran: "cpr lines";
but don't seem to be able to duplicate their examples.  I get just assembler
code (no C) even though lines.c is in my current directory, along with a message
stating "not in any module".  I can "go" and lines will execute; just can't get
to the source code like in the example.  I've also tried increasing stack
size...no help.

Anyone else try this?

    I have already found two bugs in the new Lattice compiler.
    
    1) When using the new CHIP keyword, you will get a module produced
       with invalid hunk id's.
    
    2) Under certain conditions (not yet determined) the following code
       segment will not generate correct code.
    
    	void sub_function()
        {
             <fill in some code>
        }
    
        main_function(var)
        int var;
        {
            <some code>
            sub_function();   <--- does not always generate subroutine
                                   calling code.
            <some more code>
        }
                                                
    This code did work under version 4.0 and 4.01 of Lattice.  I am
    going to send Lattice sample code tomorrow.  (I already talked to
    them.)
    
    Other than that, a couple of other programs I tried seem to work
    ok, and did generate smaller code with out the new Global Optimizer.

    I have also run into a few bugs. I had a similar problem to .-1 with
    a subroutine call. I had a call like:

		FOO (&A,&B,C)

    and the compiler generated code that pushed A on the stack twice, followed
    by C. It did NOT push B! I haven't been able to determine what made it do
    this since most of the time it does generate the proper code. By the way,
    this was easy to find using the new debugger since you can have it display
    the C source code along with the assembly code.

    I have also run into a few debugger bugs. If you set a watch point and
    open the watch point window, when you hit the watch point the contents
    of what you are watching is display in the window. However, if the window
    is not wide enough to contain the display, the data will be lobed off at
    the right. Scrolling the window will not display the rest of the data.
    You must make the window larger.

    Also, the dialogue window does not seem to support xon/xoff. This is a
    major pain if you display a large structure or something cause it scrolls
    off the screen. Booo!

    I have also had it die with a fatal error a couple of times. However, it
    is a version 1 product, and overall I think its a great debugger.

    Oh, there is one other minor pick on the compiler. I got an error 143:
    Object has no address. However, the highest documented error number is 95!

    More reviews to come. It will be interesting to see how quickly Lattice
    comes up with fixes for this stuff. The major one is the compiler
    generating bad code (ie non-working).

    Frank

Are you reporting this stuff on the Lattice BBS?

Sounds like your doing some good work finding out exactly what's wrong, let
Lattice know so they can fix it!
(In fact there may even be an update on the BBS already...)


John

    This is a minor bug...but important to those of us who like to squint!
    
    LSE has a problem with interlace mode. While in interlace mode the
    scroll bar goes to a solid bar if you go to the bottom. Once this
    happens you are forced to use the arrow keys or toggle interlace
    twice to reset it.
    
    Tom

Anyone got a file to compile from LSE yet?

I was getting problems because (I think) I as running out of memory. So I 
redefined QUAD:. Then a file that took 20 seconds or so to compile
normally (i.e. CLI) took about 5 seconds (WOW I thought!). Hoewever,
no object gets generated (I know about the compiler option -oram:)

Then I can put deliberate errors in a program, compile and get
"error: compiler reported errors". Then I F5 to see the errors
and it says "No more lc errors"???


John