GDB's #include file policy revisited

GDB's #include file policy revisited

[Mark Kettenis]

From the intefnals manual:

      With the exception of the global definitions supplied by `defs.h', a
   header file should explictily include the header declaring any
   `typedefs' et.al. it refers to.

The header `gdbtypes.h' uses `struct block *' in a few function
prototypes, which isn't declared in the file itself.  The rule above
seems to suggest that we should include `symtab.h'.  Alternatively, we
could simply use a forward declaration for `struct block *' in
`gdbtypes.h'.  That would avoid dragging in a lot of unecessary
cruft.  Some of the `tm.h' headers already do something like that, see
for example `config/i386/tm-i386.h'.  Do people agree that this is the
right thing to do?  Should I submit a patch for the internals manual
to turn this into an official policy?

Mark

Re: GDB's #include file policy revisited

[Andrew Cagney]

>>From the intefnals manual:

i take it you ment infernals :-)

> 
>       With the exception of the global definitions supplied by `defs.h', a
>    header file should explictily include the header declaring any
>    `typedefs' et.al. it refers to.
> 
> The header `gdbtypes.h' uses `struct block *' in a few function
> prototypes, which isn't declared in the file itself.  The rule above
> seems to suggest that we should include `symtab.h'.  Alternatively, we
> could simply use a forward declaration for `struct block *' in
> `gdbtypes.h'.  That would avoid dragging in a lot of unecessary
> cruft.  Some of the `tm.h' headers already do something like that, see
> for example `config/i386/tm-i386.h'.  Do people agree that this is the
> right thing to do?  Should I submit a patch for the internals manual
> to turn this into an official policy?

just adding:

	struct block;

was the intent, but it doesn't say that :-/

andrew