diff options
Diffstat (limited to 'doc')
-rw-r--r-- | doc/info/mds.texinfo | 49 |
1 files changed, 46 insertions, 3 deletions
diff --git a/doc/info/mds.texinfo b/doc/info/mds.texinfo index 62210a9..9fe6382 100644 --- a/doc/info/mds.texinfo +++ b/doc/info/mds.texinfo @@ -7800,6 +7800,19 @@ The number of slots that have been used, even if no longer used. This element is transparent to the user of the class. +@item @code{last} [@code{size_t}] +@vrindex @code{last} +@vrindex @code{hash_list_t.last} +@fnindex @code{hash_list_get} +@fnindex @code{hash_list_put} +@fnindex @code{hash_list_remove} +This variable is used for optimisation, any +time @code{hash_list_get} finds an element, its +will be stored, and it will be the first +inspected element by @code{hash_list_put} and +@code{hash_list_remove}. This element is +transparent to the user of the class. + @item @code{slots} [@code{hash_list_entry_t*}] @vrindex @code{slots} @vrindex @code{hash_list_t.slots} @@ -7864,13 +7877,15 @@ on error. Errors are however non-fatal and can be ignored, it only means that the capacity could not be reduces, because a called to @code{realloc} failed. -@item @code{hash_list_get} [(@code{const this, const KEY_T key, VALUE_T* restrict value}) @arrow{} @code{int}] +@item @code{hash_list_get} [(@code{this, const KEY_T key, VALUE_T* restrict value}) @arrow{} @code{int}] @fnindex @code{hash_list_get} Look up a value, with the key @code{key}, in the hash list @code{this}. The found value will be stored in @code{*value}. This method cannot fail, therefore it cannot return a value indicating that it failed. Instead it returns one if the entry was found and nought if it was not found. +Note that @code{this} is not specified with @code{const}; +this is because internal caching is done. @item @code{hash_list_put} [(@code{this, KEY_T key, const VALUE_T* restrict value}) @arrow{} @code{int}] @fnindex @code{hash_list_put} @@ -7954,8 +7969,8 @@ will it would return after the unmarshalling. The function shall return zero on error. @end table -@file{<libmdsserver/hash-list.h>} also defines a -macro that is uneffected by @code{CREATE_HASH_LIST_SUBCLASS}. +@file{<libmdsserver/hash-list.h>} also defines two +macros that is uneffected by @code{CREATE_HASH_LIST_SUBCLASS}. @table @asis @item @code{foreach_hash_list_entry} [(@code{hash_list_t this, size_t i, hash_list_entry_t* entry})] @@ -7992,6 +8007,34 @@ void print_hash_list(hash_list_t* table) Note thay the data type for the parameter @code{this} is not a pointer. + +@item @code{HASH_LIST_EXPECTED} [(@code{...})] +This macro is defined as @code{__builtin_expect(__VA_ARGS__, 1)}. +This means that the input value is evaluated, and may +contain the sequence-operator (comma-operator), and the +the value is expected to evaluate to @code{1}. + +It is encourage to run @code{hash_list_get} before +@code{hash_list_put} and @code{hash_list_remove}. +This way, you know exactly what is happening. +There is an optimisation in place to let +@code{hash_list_put} and @code{hash_list_remove} +skip the search for the item (unless it is the first +element) if @code{hash_list_get} was used directly +prior to @code{hash_list_put} or @code{hash_list_remove} +to find the element. This macro used to tell the +compiler that position of the element is most likely +already known but is not zero. + +If you however choose to not call @code{hash_list_get} +before @code{hash_list_put} and @code{hash_list_remove} +you should define this macro before including the header +file, but with the @code{1} changed to a @code{0}. If +you on the other hand do not know if you are going to +call @code{hash_list_get} before @code{hash_list_put} +and @code{hash_list_remove} you should define it to +expand to the input verbatim, that is, have the value +@code{__VA_ARGS__}. @end table |