Re: A "Via Pascal" Modula 2 Compiler ?

>>> On 1/25/2007 at 7:07 PM, Andreas F.
Borchert<comp.lang.modula2@[EMAIL PROTECTED]
>

wrote:
> On 2007-01-25, Marco van de Voort <marcov@[EMAIL PROTECTED]
> wrote:
>> What is a symbol file btw? I guess some per module stored info,
>> but what is stored exactly?
> 
> A symbol file is a compiler-generated representation of a definition
> module including a unique key and the keys of the im****ted modules.
> The term »symbol file« was coined by the developers of the Modula-2
> compilers at ETH Zürich. An EBNF syntax of the symbol file format can
> be found at http://www.mathematik.uni-ulm.de/modula/ipf/s1.
> 
> The im****tant point is that definition modules are processed separately,
> i.e. you can generate the symbol file even if the associated
> implementation module does not yet exist. And, if an implementation
> module is to be compiled, its associated symbol file must exist like
> the symbol files of the closure of all im****ted modules.

Another interesting thing in the M68K compiler(s) from ETH is that the
SYM file is designed to minimize the number of SYM files that might have
to be opened during a compile by including in it all the definitions from
im****ted other modules on which it depends, along with those module names
and keystamps.

eg, if you write a DEF module that im****ts one TYPE from Windows.def, that
type
definition is included in the SYM file of your DEF module. This makes your
SYM file a little bigger, but can avoid the need to open and read
Windows.SYM
just to compile your module(s).

There is a doc from the ETH archive explaining the technique in detail; I
can get
a reference to it if anyone is interested.

It works well, though in very rare instances I think I have detected a
subtle bug
(resulting in an erroneous "double definition error" or maybe expression
incompatibility
error). I guess it also means that some module inconsistencies are not
detected until
link time, but in practice that has not been a problem since the "deeper"
SYM files
change little in comparison to the shallower ones.

Tom