Question about a cast

[Richard Harter]

I have a couple of questions about the following code fragment:

struct list_info {
void * first;
void * last;
};

struct pipe_data_list {
void * data;
struct agent_port_list * dest;
struct pipe_data_list * link;
};
....

struct list_info next = {0};
struct list_info processed;
....

/* This code fragment concatenates lists next and processed */

if (next.last) {
((struct pipe_data_list *)next.last)->link = processed.first ;
next.last = processed.last;
} else next = processed;

What is going on here is that list_info is a generic container for null
terminated lists that records pointer to the first and last nodes of the
contained list. A reason for doing this is that it enables concatenating
lists without scanning them. The questions are about this line:

((struct pipe_data_list *)next.last)->link = processed.first ;

which compiles without warning. Are there any gotchas that I've
overlooked?

Richard Harter, cr*@tiac.net
http://home.tiac.net/~cri, http://www.varinoma.com
Save the Earth now!!
It's the only planet with chocolate.

[vippstar]

On Sep 23, 7:51 pm, [e-mail] (Richard Harter) wrote:

I have a couple of questions about the following code fragment:

struct list_info {
void * first;
void * last;
};

struct pipe_data_list {
void * data;
struct agent_port_list * dest;
struct pipe_data_list * link;
};
....

struct list_info next = {0};
struct list_info processed;
....

/* This code fragment concatenates lists next and processed */

if (next.last) {
((struct pipe_data_list *)next.last)->link = processed.first ;
next.last = processed.last;
} else next = processed;

What is going on here is that list_info is a generic container for null
terminated lists that records pointer to the first and last nodes of the
contained list. A reason for doing this is that it enables concatenating
lists without scanning them. The questions are about this line:

((struct pipe_data_list *)next.last)->link = processed.first ;

which compiles without warning. Are there any gotchas that I've
overlooked?

No, why do you think so? (assuming next.last indeed does point to a
struct pipe_data_list or a compatible struct)

[Richard Heathfield]

Richard Harter said:

>
I have a couple of questions about the following code fragment:

struct list_info {
void * first;
void * last;
};

struct pipe_data_list {
void * data;
struct agent_port_list * dest;
struct pipe_data_list * link;
};
....

struct list_info next = {0};
struct list_info processed;
....

/* This code fragment concatenates lists next and processed */

if (next.last) {
((struct pipe_data_list *)next.last)->link = processed.first ;
next.last = processed.last;
} else next = processed;

What is going on here is that list_info is a generic container for null
terminated lists that records pointer to the first and last nodes of the
contained list. A reason for doing this is that it enables concatenating
lists without scanning them. The questions are about this line:

((struct pipe_data_list *)next.last)->link = processed.first ;

which compiles without warning. Are there any gotchas that I've
overlooked?

As far as I can see, there is only one problem with the code, which is that
it causes someone as bright as Richard Harter to question whether it is
correct. Heaven knows what a maintenance weenie would make of it.

struct list_info *tmp = next.last;
tmp->link = processed.first ;

Don't you think that looks neater? YMMV, of course.

--
Richard Heathfield <http://www.cpax.org.uk >
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999

[Richard Harter]

On Tue, 23 Sep 2008 17:23:20 +0000, Richard Heathfield
<rj*@see.sig.in validwrote:

>Richard Harter said:

>>
I have a couple of questions about the following code fragment:

struct list_info {
void * first;
void * last;
};

struct pipe_data_list {
void * data;
struct agent_port_list * dest;
struct pipe_data_list * link;
};
....

struct list_info next = {0};
struct list_info processed;
....

/* This code fragment concatenates lists next and processed */

if (next.last) {
((struct pipe_data_list *)next.last)->link = processed.first ;
next.last = processed.last;
} else next = processed;

What is going on here is that list_info is a generic container for null
terminated lists that records pointer to the first and last nodes of the
contained list. A reason for doing this is that it enables concatenating
lists without scanning them. The questions are about this line:

((struct pipe_data_list *)next.last)->link = processed.first ;

which compiles without warning. Are there any gotchas that I've
overlooked?

As far as I can see, there is only one problem with the code, which is that
it causes someone as bright as Richard Harter to question whether it is
correct. Heaven knows what a maintenance weenie would make of it.

My view is that it is good for maintenance weenies; they need the
mental exercise. :-)

>
struct list_info *tmp = next.last;
tmp->link = processed.first ;

Don't you think that looks neater? YMMV, of course.

One minor problem - it's wrong. That should have been
struct pipe_data_list *tmp = next.last;

I dunno about neatness. I look at the posted form and say to
myself, there sure is a lot of crap on the LHS. OTOH I don't
much like sticking declarations in the middle of code. Yes, in
the posted code it slides in nicely because there is that test on
next.last and a brace immediately after it. Suppose I hadn't had
the test there? Then there wouldn't have been a natural reason
for braces to be there. Putting a pair in so that one could put
in a declaration for a temp for two lines of code would be rather
hokey. Actually, when I looked at your version, my first thought
was "Is this some kind of C99 crap?" and then it dawned on me
that it was a declaration with an initialization. Being old
fashioned, I like my declarations up front, gathered together,
with nice documentation.

Still and all, I suppose I will do it your way - I don't like
long winded statements.

Richard Harter, cr*@tiac.net
http://home.tiac.net/~cri, http://www.varinoma.com
Save the Earth now!!
It's the only planet with chocolate.

[Richard Heathfield]

Richard Harter said:

On Tue, 23 Sep 2008 17:23:20 +0000, Richard Heathfield
<rj*@see.sig.in validwrote:

<snip>

>>
struct list_info *tmp = next.last;
tmp->link = processed.first ;

Don't you think that looks neater? YMMV, of course.

One minor problem - it's wrong. That should have been
struct pipe_data_list *tmp = next.last;

Right. Did you see how easy my version was to debug? :-)

I dunno about neatness. I look at the posted form and say to
myself, there sure is a lot of crap on the LHS. OTOH I don't
much like sticking declarations in the middle of code. Yes, in
the posted code it slides in nicely because there is that test on
next.last and a brace immediately after it. Suppose I hadn't had
the test there? Then there wouldn't have been a natural reason
for braces to be there.

And so it would have had to be stuck in up top somewhere, yes. Not quite so
sweet then.

Putting a pair in so that one could put
in a declaration for a temp for two lines of code would be rather
hokey. Actually, when I looked at your version, my first thought
was "Is this some kind of C99 crap?"

Um, you clearly don't know me very well, then. :-)

and then it dawned on me
that it was a declaration with an initialization. Being old
fashioned, I like my declarations up front, gathered together,
with nice documentation.

Docs cost extra, payable in advance.

Still and all, I suppose I will do it your way - I don't like
long winded statements.

Never run for office.

--
Richard Heathfield <http://www.cpax.org.uk >
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999