redis sorted set source code analysis

 sorted set in t_zset.c file。

Insert operation:

zaddCommand -----> zaddGenericCommand

{  zzlFind

{  zzlDelete

{  zzlInsert

-------- { ziplist : http://blog.nosqlfan.com/html/3919.html

-------- { zzlGetScore

-------- { zzlInsertAt

--------------- { ziplistPush

--------------- { ziplistInsert

{ dictAdd

{ dictFind

{ dictGetKey

{ dictGetVal

{  zslInsert : zskiplist Insert

------------:   http://www.cnblogs.com/xuqiang/archive/2011/05/22/2053516.html

{ zsetConvert : if the element is too large or the list becomes too long executing zzlInsert

-------------- { dictCreate

-------------- { zslCreate
 

#############################################################

args:

@ zs -> dict : 

---------------:  zs -----> zset struct

@ ele -> redisObject

@ &znode -> score

@ elements : add element number

@ redisObject :  http://tech.it168.com/a2011/0818/1234/000001234478_all.shtml

@ zobj is redisObject 

-------- zobj->encoding is the type of sort set key
 #############################################################

Program process:

1. Parsing all the scores

2. lookup the key exist ?

---- if not exist

-------- if has not ziplist or ziplist value < len

------------creatzsetobject

-------- else createzsetziplistobject

-------- dbadd key

3. Insert the key

---- for to insert all element

---- if the node is less the max-ziplist-entries or node value less max-ziplist-value

-------- use ziplist struct 

--------  else use skiplist

---- else if the key already encoding in REDIS_ENCODING_SKINGLIST

-------- insert the node in skiplist and dict

##############################################################

Summary:

1. the time complexity is log(n) because of skiplist

2. the time of dict is 0(1). use mapping like: key ----> score