[FEATURE REQUEST + PATCH] utstring_clone()

[silvioprog]

Hi.

It would be nice to provide a macro for list cloning, something like this code below:

#ifdef __GNUC__

#define utstring_clone(src)               \
({                                        \
  UT_string *dst = NULL;                  \
  if ((src)->d != NULL) {                 \
    utstring_new(dst);                    \
    (dst)->n = (src)->n;                  \
    (dst)->i = (src)->i;                  \
    memcpy((dst)->d, (src)->d, (src)->n); \
  }                                       \
  dst;                                    \
})

#else

UT_string *utstring_clone(UT_string *src) {
  UT_string *dst = NULL;
  if (src->d != NULL) {
    utstring_new(dst);
    dst->n = src->n;
    dst->i = src->i;
    memcpy(dst->d, src->d, src->n);
  }
  return dst;
}

#endif

What do you thing? If you agree this code I can create a pull request. :-)

Thank you!

[Quuxplusone]

I would think that a more consistent definition for utstring_clone would be simply

#define utstring_clone(dst,src) \
do {                            \
  utstring_clear(dst);          \
  utstring_concat(dst, src);    \
} while (0)

Would such a macro be useful to you, or would it be useful only if it returned the value?

I'd be a little worried about having utstring_clone return a value, not just because it would be inconsistent with the rest of the library, but also because it seems like that would encourage people to write

UT_string *foo; utstring_new(foo); utstring_bincpy("foo", 3);
some_client_function(foo, utstring_clone(foo));
utstring_free(foo);

and forget to utstring_free the anonymous cloned string.

[silvioprog]

#define utstring_clone(dst,src) \
do {                            \
  utstring_clear(dst);          \
  utstring_concat(dst, src);    \
} while (0)

Awesome definition, following the same utstring_concat(dst,src) signature. It may be useful because the "cloning" feature is very common in many programming languages. :-)

I have a similar question, but regarding uthash.h. Would it be possible to have a "clone" feature too?

[Quuxplusone]

For uthash.h, you'd have to be more specific about what you want this "clone" operation to do. I imagine that you'd be looking for HASH_SELECT, which already exists.