home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
EnigmA Amiga Run 1997 July
/
EnigmA AMIGA RUN 20 (1997)(G.R. Edizioni)(IT)[!][issue 1997-07 & 08][EAR-CD IV].iso
/
earcd
/
biz
/
haage
/
sc_pch4.lha
/
STC-History-Engl
< prev
next >
Wrap
Text File
|
1996-09-27
|
3KB
|
92 lines
10.08.96
* Wrong data width at assign of "enum" constants to Bytes or Words
11.08.96 (V 1.10.02)
* Up form optimisation level 5 e.g. triangle exchanges are making
trouble when the help variable is at a register and the both other
variables are at the stack.
* The "mathieeedoubbas.library" has not been closed by the compiler.
* Constant __STORM__ now has a value of 11
18.08.96 (V 1.10.03)
* Initialisation of static data: unnecessary INIT code was created
for pointers to functions.
19.08.96
* If there was a switch instruction with a difference between the
smallest and biggest case label >=231, then there an attempt to
create a jump table. That led to a Workspace overflow.
31.08.96
* 68k Code generation fault (> optimisation level 5) removed.
01.09.96 (V 1.10.04)
* If there was a vector of objects created with "new[number]"
it could happen that the constructor was called for the 1st element
of the vector only. And this was done as often as the vector has
elements.
* When a function used more than 32KByte of Stack frame it came very
often to an incorrect addressing.
05.09.96
* When a function gives back an object that will be given directly as
an argument to another function this caused wrong code or compiler
panic.
09.09.96
* There was an Enforcer hit when the error message "Unexpected end
of source" appeared.
* There was a crash when there was a defined, but not declared
destructor and debug-file-creation was active.
10.09.96
* There was wrong code when there was an access to an array of
"double" (without FPU) or "long long": When an element of an
array was loaded in double data register and at the same time the
index expression stands in the first of these two data registers,
then the first long word was loaded first, the index expression was
trashed and then the second long word was loaded with a disallowed
index.
12.09.96 (V 1.10.05)
* From 1.10.03 on there was an optimisation test in the compiler
that should not have been there - sorry for that.
17.09.96 (V 1.10.06)
* The "double" comparison "<" and ">=" were handled as communicative,
for that the compiler exchanged their arguments if this seamed to
be of any advantage.
* When there was register confusion before a double arithmetic
library function it could happen principally that he both arguments
were exchanged first, but were not swapped back later.
21.09.96 (V 1.10.07)
* There was a compiler crash or Enforcer hit, when the conditioned
operator "?:" was abused with two "void" expressions instead of
an "if" instruction.
* When there was something derived from a template class, the name
of the basis class, e.g. "Basis<int>", could not be used in the
initialisation list.
* Access to bit fields was very faulty.