naive GDB programming style questions

naive GDB programming style questions

[David Carlton]

1) It seems to me that some parts of GDB's source use NULL while other
   parts use 0.  Is one or the other of these generally preferred?

2) Am I correct in observing that GDB frowns on code like the
   following:

     char *p = calculate_p ();

     if (!p)
       p_is_zero ();

   preferring this instead?

     char *p = calculate_p ();

     if (p == NULL)
       p_is_zero ();

   Does it matter that p in a pointer rather than an integer, or that
   the code is testing for zeroness rather than nonzeroness?

3) Is it possible to get CC Mode to indent in the way that GDB seems
   to prefer?  I'm having a hard time getting structs to be indented
   as follows:

     struct foo
       {
         int mem;
       };

   without screwing up my preferred indentation when doing non-GDB
   programming, namely

     struct foo {
       int mem;
     };

   Of course, I can write functions to toggle between the two; but
   given that both styles seem to be able to coexist for enums, maybe
   it's possible to get them to coexist for enums.  (I'm certainly no
   CC Mode expert...)

David Carlton
carlton@math.stanford.edu

Re: naive GDB programming style questions

[Andrew Cagney]

> 1) It seems to me that some parts of GDB's source use NULL while other
>    parts use 0.  Is one or the other of these generally preferred?
> 
> 2) Am I correct in observing that GDB frowns on code like the
>    following:
> 
>      char *p = calculate_p ();
> 
>      if (!p)
>        p_is_zero ();

I'll assume that you ment ``p_is_null()''.

>    preferring this instead?
> 
>      char *p = calculate_p ();
> 
>      if (p == NULL)
>        p_is_zero ();
> 
>    Does it matter that p in a pointer rather than an integer, or that
>    the code is testing for zeroness rather than nonzeroness?

If you've a copy of the ISO C and C++ manuals, have a look at what they 
have to say about ``NULL'' pointers.  It's weird.

Anyway, for GDB, ``p == NULL'' is recommended to make it clear that the 
pointer is being tested and not the underlying value.

> 3) Is it possible to get CC Mode to indent in the way that GDB seems
>    to prefer?  I'm having a hard time getting structs to be indented
>    as follows:
> 
>      struct foo
>        {
>          int mem;
>        };

Just use:

struct foo
{
   int mem;
};

which is what is output by gdb_indent.sh and emacs.  The strange 
indentation that you've encountered dates back to an earlier version of 
indent.

Must get around to sending the FSF coding standards group a patch that 
documents this (it was agreed to in principle some time ago).

>    without screwing up my preferred indentation when doing non-GDB
>    programming, namely
> 
>      struct foo {
>        int mem;
>      };
> 
>    Of course, I can write functions to toggle between the two; but
>    given that both styles seem to be able to coexist for enums, maybe
>    it's possible to get them to coexist for enums.  (I'm certainly no
>    CC Mode expert...)

Andrew

Re: naive GDB programming style questions

[David Carlton]

On Mon, 09 Sep 2002 21:45:54 -0400, Andrew Cagney <ac131313@ges.redhat.com> said:

> I'll assume that you ment ``p_is_null()''.

Right.

> If you've a copy of the ISO C and C++ manuals, have a look at what
> they have to say about ``NULL'' pointers.  It's weird.

Really?  Hmm.  I know that NULL doesn't work as well in C++ as it does
in C, so I've gotten used to using 0 there.

Though if you're referring to the fact that NULL's underlying bit
representation might not be 0, I'm not sure that's a big deal here.

> Anyway, for GDB, ``p == NULL'' is recommended to make it clear that
> the pointer is being tested and not the underlying value.

Okay, then that's what I'll do.

> Just use:

> struct foo
> {
>    int mem;
> };

> which is what is output by gdb_indent.sh and emacs.

Excellent.

David

Re: naive GDB programming style questions

[Tom Tromey]

>>>>> "David" == David Carlton <carlton@math.stanford.edu> writes:

David> 3) Is it possible to get CC Mode to indent in the way that GDB seems
David>    to prefer?  I'm having a hard time getting structs to be indented
David>    as follows:
David>    without screwing up my preferred indentation when doing non-GDB
David>    programming, namely

Change your settings depending on the file you are editing.

Here is a piece of code I used for this purpose a long time ago.  For
C source files in a particular directory, it uses the Tcl style (a
variant of the BSD style); for other files it uses GNU.  Rewrite as
appropriate and attach to your c-mode-hook.

(defun tjt-set-c-style ()
  (if (and (buffer-file-name)
	   (string-match "/home/syzygy/tclstuff" (buffer-file-name)))
      (progn
	(c-set-style "BSD")
	(make-local-variable 'c-basic-offset)
	(setq c-basic-offset 4))
    (c-set-style "GNU")))

There are probably lots of other ways to accomplish the same thing.
And of course you can set anything there; for instance the behavior of
electric braces (that sounds like what you want).

Tom

Re: naive GDB programming style questions

[David Carlton]

On 09 Sep 2002 20:52:14 -0700, David Carlton <carlton@math.Stanford.EDU> said:
> On Mon, 09 Sep 2002 21:45:54 -0400, Andrew Cagney <ac131313@ges.redhat.com> said:

>> If you've a copy of the ISO C and C++ manuals, have a look at what
>> they have to say about ``NULL'' pointers.  It's weird.

> Though if you're referring to the fact that NULL's underlying bit
> representation might not be 0, I'm not sure that's a big deal here.

Actually, that raises another question: if I'm allocating an array of
pointers that I want initialized to NULL, am I allowed to use
xcalloc() to handle that, or do I have to loop through the memory
myself to set all the pointers to NULL?  Because the former is
technically incorrect, but I don't know if GDB runs on any platforms
for which it wouldn't work; I think I've seen parts of GDB initialize
pointers to NULL using xcalloc or memset, but maybe I'm wrong.

David Carlton
carlton@math.stanford.edu

Re: naive GDB programming style questions

[Andrew Cagney]

> On 09 Sep 2002 20:52:14 -0700, David Carlton <carlton@math.Stanford.EDU> said:
> 
>> On Mon, 09 Sep 2002 21:45:54 -0400, Andrew Cagney <ac131313@ges.redhat.com> said:
> 
> 
>>> If you've a copy of the ISO C and C++ manuals, have a look at what
>>> they have to say about ``NULL'' pointers.  It's weird.
> 
> 
>> Though if you're referring to the fact that NULL's underlying bit
>> representation might not be 0, I'm not sure that's a big deal here.
> 
> 
> Actually, that raises another question: if I'm allocating an array of
> pointers that I want initialized to NULL, am I allowed to use
> xcalloc() to handle that, or do I have to loop through the memory
> myself to set all the pointers to NULL?  Because the former is
> technically incorrect, but I don't know if GDB runs on any platforms
> for which it wouldn't work; I think I've seen parts of GDB initialize
> pointers to NULL using xcalloc or memset, but maybe I'm wrong.

You can use memset() and xcalloc().

Andrew

Re: naive GDB programming style questions

[Andrew Cagney]

PS: If you encounter a file that has the old ``dud'' struct style 
indentation then running it (otherwize unchanged) through gdb_indent.sh 
comes under the ``obvious fix rule''.  It lets you get any indentation 
issues out of the way up before you start making changes.

Andrew


> On Mon, 09 Sep 2002 21:45:54 -0400, Andrew Cagney <ac131313@ges.redhat.com> said:
> 
> 
>> I'll assume that you ment ``p_is_null()''.
> 
> 
> Right.
> 
> 
>> If you've a copy of the ISO C and C++ manuals, have a look at what
>> they have to say about ``NULL'' pointers.  It's weird.
> 
> 
> Really?  Hmm.  I know that NULL doesn't work as well in C++ as it does
> in C, so I've gotten used to using 0 there.
> 
> Though if you're referring to the fact that NULL's underlying bit
> representation might not be 0, I'm not sure that's a big deal here.
> 
> 
>> Anyway, for GDB, ``p == NULL'' is recommended to make it clear that
>> the pointer is being tested and not the underlying value.
> 
> 
> Okay, then that's what I'll do.
> 
> 
>> Just use:
> 
> 
>> struct foo
>> {
>>    int mem;
>> };
> 
> 
>> which is what is output by gdb_indent.sh and emacs.
> 
> 
> Excellent.
> 
> David
>