[comp.sys.amiga.tech] Manx bugs

paolucci@snll-arpagw.UUCP (Sam Paolucci) (11/26/88)

I just thought you might want to know of a patch for the compiler I
got from Manx which fixes the bug in a long switch statement containing
an 'x' or 'X' case.  The bug manifested itself by generating bad code
for the assembler.  Well here it is:

main()
{
	int fd;

	fd = open("cc", 2);
	lseek(fd,0x114E9L, 0);
	write(fd, "\x37", 1);
	close(fd);
}

Manx told me that this just changes a byte from 0x38 to 0x37 at the
above offset, but my 3.6a cc says that the byte before the patch is
applied is 0x36, not 0x38.  Anyway, I applied the patch and I tried it 
on a code which displayed the bug, and now seems to compile fine.

You might also want to know that it looks like the many bugs I reported, 
as well as the ones that others have reported, won't get fixed until
4.1 appears.  And they tell me that they plan to release 4.1 in early
March of next year.  That is, unless enough pressure is put on them
to release patches for major ones before then.
-- 
					-+= SAM =+-
"the best things in life are free"

				ARPA: paolucci@snll-arpagw.llnl.gov

andrew@bhpese.bhpese.oz (Andrew Steele) (12/01/88)

Talking of Manx Bugs two I've found are the following :

1.	/* Minimal Example */
	main()
	{
		int i;
		i++ %= 3;
	}

	Whilst I now know that this should produce an error because you
	can't increment an lvalue.  What it shouldn't do is cause the 
	compiler to generate a guru which it does!!

2.	This actually applies to SDB or possibly the OS.
	When sdb starts up it draws the window it works in, then goes and
	loads in your source program and then displays the file in the 
	window.  If you try to move the window after it has come up but
	before the source file is displayed, the window can be made to 
	wrap on the screen if you still have the left mousebutton held down 
	while sdb tries to update its window to display the source file.
	i.e.
		|------------------| <-- Screen
		|   a----------b   |
		|   |          | <-|---- SDB Window
		|   c----------d   |
		|------------------|

		Holding the Left Mousebutton Down to keep the window
		selected during the window update as the file is displayed,
		by moving the mouse you can produce :

		|------------------| <-- Screen
		|-----d     c------|
		|-----b     a------|
		|     |     |    x-|---- Wrapped Window
		|------------------|
				   _____
Andrew Steele 			  /_   _\    Spengat Technologies, 	
			 	  _ | | _    c/o Electrical Computer Services,
ACSnet  : andrew@bhpese.oz	 /__| |__\   BHP Rod & Bar Products Division,
INTERNET: andrew@bhpese.oz.au	             Newcastle, NSW, Australia.
UUCP    : ...!{uunet,mcvax}!munnari!bhpese.oz!andrew

dat01@butterix.liu.se (Dat) (12/05/88)

In article <154@bhpese.bhpese.oz> andrew@bhpese.bhpese.oz (Andrew Steele) writes:

>		Holding the Left Mousebutton Down to keep the window
>		selected during the window update as the file is displayed,
>		by moving the mouse you can produce :
>
>		|------------------| <-- Screen
>		|-----d     c------|
>		|-----b     a------|
>		|     |     |    x-|---- Wrapped Window
>		|------------------|

 You can acheive a similar effect if you hold down left amiga-n (or is it m?)
bringing the Workbench screen to the front and, while holding the keys down,
drag a window. They say the Amiga has no limits :-)
 This has obviously nothing to do with Manx.
_____________                                                              ///
WWWWW  Mikael Karlsson, Lovsattersvagen 10, S-585 98  LINKOPING, SWEDEN   ///
{@v@}  "I may be artificial, but I'm not stupid."                     \\\///
 \w/   "Give me a fixed address, and I will move the World"            \XX/

mike@ames.arc.nasa.gov (Mike Smithwick) (04/07/90)

[]

I've been having some more problems with Manx (5.0b).

The "-c2" compiler option for me produced bad assembly code that the
assembler barfed on. Anyone have the same problem?

The manual claims that with the ffp math library (-ff option) I can
get 64 bit floats by using "long double". Ehhhhh! I couldn't tell any
difference. Does it work?

When using the -f8 option (generates 68881  code). fscanf breaks. I cannot
read in float values, using %f of %lf. 

Anyone have any answers? Solutions? Good jokes?

mike



                                                      *** mike smithwick ***
"I'm totally awed by what you've done!" (Arthur C. Clarke 
 commenting about Distant Suns)
[disclaimer : nope, I don't work for NASA, I take full blame for my ideas]