Replace char * with something sensible

Replace char * with something sensible

[Eli Zaretskii]

Rumor has it that latest versions of GCC whine about mixing "char *"
and "unsigned char *".  To fix that, Andrew committed some patches
that replace these with a "bfd_byte *", but I think we should not use
BFD data types in GDB unless they refer to data structures returned by
functions from the BFD library.

Possible solutions:

 . use "void *" (suggested by Mark Kettenis).

 . define a new data type gdb_byte and use "gdb_byte *".

I think Mark's suggestion should be considered first, since it avoids
the pain of introducing yet another custom data type.

Re: Replace char * with something sensible

[Daniel Jacobowitz]

On Tue, May 03, 2005 at 10:51:02PM +0300, Eli Zaretskii wrote:
> Rumor has it that latest versions of GCC whine about mixing "char *"
> and "unsigned char *".  To fix that, Andrew committed some patches
> that replace these with a "bfd_byte *", but I think we should not use
> BFD data types in GDB unless they refer to data structures returned by
> functions from the BFD library.
> 
> Possible solutions:
> 
>  . use "void *" (suggested by Mark Kettenis).
> 
>  . define a new data type gdb_byte and use "gdb_byte *".
> 
> I think Mark's suggestion should be considered first, since it avoids
> the pain of introducing yet another custom data type.

I am marginally in favor of gdb_byte.  What pain would it cause,
besides a huge sed for the existing bfd_byte cases?

These are byte-oriented buffers, so using a type where we can perform
byte-oriented arithmetic without superfluous casts seems like a good
choice to me.  Converting to void * (and not using the GCC extension
which allows arithmetic on void *) would be a painful process.

It's not as if non-8-bit-byte support is anywhere on our horizon.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Replace char * with something sensible

[Mark Kettenis]

   Date: Tue, 3 May 2005 15:58:15 -0400
   From: Daniel Jacobowitz <drow@false.org>

   On Tue, May 03, 2005 at 10:51:02PM +0300, Eli Zaretskii wrote:
   > Rumor has it that latest versions of GCC whine about mixing "char *"
   > and "unsigned char *".  To fix that, Andrew committed some patches
   > that replace these with a "bfd_byte *", but I think we should not use
   > BFD data types in GDB unless they refer to data structures returned by
   > functions from the BFD library.
   > 
   > Possible solutions:
   > 
   >  . use "void *" (suggested by Mark Kettenis).
   > 
   >  . define a new data type gdb_byte and use "gdb_byte *".
   > 
   > I think Mark's suggestion should be considered first, since it avoids
   > the pain of introducing yet another custom data type.

   I am marginally in favor of gdb_byte.  What pain would it cause,
   besides a huge sed for the existing bfd_byte cases?

I hope my (somewhat lengthy) post can change your mind ;-).

   These are byte-oriented buffers, so using a type where we can perform
   byte-oriented arithmetic without superfluous casts seems like a good
   choice to me.  Converting to void * (and not using the GCC extension
   which allows arithmetic on void *) would be a painful process.

We have -Wpointer-arith in our standard warning flags so there's no
risk in us using that darn GCC extension.

   It's not as if non-8-bit-byte support is anywhere on our horizon.

And it is probably completely irrelevant for this discussion.

Mark