{"id":111063,"date":"2025-04-10T07:00:00","date_gmt":"2025-04-10T14:00:00","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/oldnewthing\/?p=111063"},"modified":"2025-04-10T08:45:54","modified_gmt":"2025-04-10T15:45:54","slug":"20250410-00","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/oldnewthing\/20250410-00\/?p=111063","title":{"rendered":"Function overloading is more flexible (and more convenient) than template function specialization"},"content":{"rendered":"

A colleague of mine was having trouble specializing a templated function. Here’s a simplified version.<\/p>\n

template<typename T, typename U>\r\nbool same_name(T const& t, U const& u)\r\n{\r\n    return t.name() == u.name();\r\n}\r\n<\/pre>\n
They wanted to provide a specialization for the case that the parameters are a Widget and a string literal.<\/p>\n
template<>\r\nbool same_name<Widget, const char[]>(Widget const& widget, const char name[])\r\n{\r\n    return strcmp(widget.descriptor().name().c_str(), name) == 0;\r\n}\r\n<\/pre>\n
However, this failed to compile:<\/p>\n
\/\/ msvc\r\nerror C2912: explicit specialization 'bool same_name<Widget,const char[]>(const Widget &,const char [])' is not a specialization of a function template\r\n<\/pre>\n
What do you mean “not a specialization of a function template”? I mean doesn’t it look like a specialization of a function template? It sure follows the correct syntax for a function template specialization.<\/p>\n
The error message from gcc is a little more helpful:<\/p>\n
error: template-id 'same_name<Widget, const char []>' for 'bool same_name(const Widget&, const char*)' does not match any template declaration\r\n<\/pre>\n
Okay, so gcc recognized that it’s a specialization of a function template, but it couldn’t find a match. What is this “match” talking about?<\/p>\n
The error message from clang helps even more:<\/p>\n
error: no function template matches function template specialization 'same_name'\r\n| bool same_name<Widget, const char[]>(Widget const& widget, const char name[])\r\n\r\nnote: candidate template ignored: could not match 'bool (const Widget &, const const char (&)[])' against 'bool (const Widget &, const char *)'\r\n<\/pre>\n
Okay, now we’re getting somewhere. The compiler is taking the specialization we provided and is unable to match it against the non-specialized version.<\/p>\n
And that’s where we see the problem. If we substitute Widget<\/code> and const char[]<\/code> into the original declaration of bool same_name(T const& t, U const& u)<\/code>, we get<\/p>\n
    bool same_name(Widget const& t, const char(& u)[]);\r\n<\/pre>\n
But this isn’t the function signature of our proposed specialization. Our proposed specialization takes a const char*<\/code> as the final parameter, since function and array parameters in parameter lists are rewritten as pointers: [dcl.fct]<\/b>(4)<\/a>: “any parameter of type ‘array of T<\/tt>‘ or of function type T is adjusted to be ‘pointer to T<\/tt>‘.”<\/p>\n
That’s what msvc was trying to tell us when it said “is not a specialization of a function template”: “What you wrote sure looks like a specialization of a function template, but it’s not because the signature is wrong.” Perhaps a better message would be “is not a valid<\/i> specialization of a function template” or “does not correspond to<\/i> a specialization of a function template.”<\/p>\n
A valid specialization would be<\/p>\n
template<>\r\nbool same_name<Widget, const char*>(Widget const& widget, const char *const& name)\r\n{\r\n    return strcmp(widget.descriptor().name().c_str(), name) == 0;\r\n}\r\n<\/pre>\n
That sure looks clunky, but it doesn’t have to be.<\/p>\n
You don’t need to do specialization at all: You can use overloading.<\/p>\n
bool same_name(Widget const& widget, const char* name)\r\n{\r\n    return strcmp(widget.descriptor().name().c_str(), name) == 0;\r\n}\r\n<\/pre>\n
The nice thing about overloading is that you don’t have to be a perfect match for the original template. Here, we take the second parameter by value instead of by reference. You can even change the return value in an overload.<\/p>\n
std::optional<bool> same_name(Widget const& widget, const char* name)\r\n{\r\n    if (!widget.is_name_known()) return std::nullopt;\r\n    return strcmp(widget.descriptor().name().c_str(), name) == 0;\r\n}\r\n<\/pre>\n
In this case, we change the return type from bool<\/code> to std::optional<bool><\/code> to be able to express the “I don’t know” case.<\/p>\n
Function templates cannot be partially specialized, but that’s okay: You can get the same effect via overloading.<\/p>\n
template<typename U>\r\nstd::optional<bool> same_name(Widget const& widget, U const& u)\r\n{\r\n    if (!widget.is_name_known()) return std::nullopt;\r\n    return widget.name() == u.name();\r\n}\r\n<\/pre>\n\n\n\n\n\n\n\n
\u00a0<\/th>\nClass<\/th>\nFunction<\/th>\n<\/tr>\n
Can template<\/th>\nYes<\/td>\nYes<\/td>\n<\/tr>\n
Can specialize<\/th>\nYes<\/td>\nYes<\/td>\n<\/tr>\n
Can partially specialize<\/th>\nYes<\/td>\nNo<\/td>\n<\/tr>\n
Can overload<\/th>\nNo<\/td>\nYes<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Template functions: Don’t specialize them. Overload them.<\/p>\n","protected":false},"excerpt":{"rendered":"
You can change more things in an overload.<\/p>\n","protected":false},"author":1069,"featured_media":111744,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[25],"class_list":["post-111063","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-oldnewthing","tag-code"],"acf":[],"blog_post_summary":"
You can change more things in an overload.<\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/111063","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/users\/1069"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/comments?post=111063"}],"version-history":[{"count":0,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/111063\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media\/111744"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media?parent=111063"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/categories?post=111063"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/tags?post=111063"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}