ÓÑÇéÌáʾ£ºÈç¹û±¾ÍøÒ³´ò¿ªÌ«Âý»òÏÔʾ²»ÍêÕû£¬Çë³¢ÊÔÊó±êÓÒ¼ü¡°Ë¢Ð¡±±¾ÍøÒ³£¡
VB2008´ÓÈëÃŵ½¾«Í¨(PDF¸ñʽӢÎÄ°æ)-µÚ28²¿·Ö
¿ì½Ý²Ù×÷: °´¼üÅÌÉÏ·½Ïò¼ü ¡û »ò ¡ú ¿É¿ìËÙÉÏÏ·ҳ °´¼üÅÌÉ쵀 Enter ¼ü¿É»Øµ½±¾ÊéĿ¼ҳ °´¼üÅÌÉÏ·½Ïò¼ü ¡ü ¿É»Øµ½±¾Ò³¶¥²¿! Èç¹û±¾ÊéûÓÐÔĶÁÍ꣬ÏëÏ´μÌÐø½Ó×ÅÔĶÁ£¬¿ÉʹÓÃÉÏ·½ "Êղص½ÎÒµÄä¯ÀÀÆ÷" ¹¦ÄÜ ºÍ "¼ÓÈëÊéÇ©" ¹¦ÄÜ£¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡search¡¡algorithm¡¡that¡¡uses¡¡the¡¡Node¡¡type¡¡does¡¡not¡¡need¡¡to¡¡be¡¡changed¡£¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡öNote¡¡¡¡When¡¡you¡¡have¡¡the¡¡ability¡¡to¡¡create¡¡code¡¡that¡¡localizes¡¡changes¡¡without¡¡affecting¡¡other¡¡pieces¡¡of¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡code£»¡¡it¡¡is¡¡called¡¡decoupling¡¡code¡£¡¡You¡¡want¡¡to¡¡write¡¡code¡¡that¡¡is¡¡decoupled¡¡from¡¡other¡¡code£»¡¡so¡¡that¡¡when¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡changes¡¡are¡¡made¡¡in¡¡one¡¡piece¡¡of¡¡code£»¡¡other¡¡pieces¡¡of¡¡code¡¡continue¡¡functioning¡£¡¡As¡¡you¡¡will¡¡experience¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡when¡¡developing¡¡code£»¡¡decoupling¡¡of¡¡code¡¡is¡¡a¡¡daily¡¡struggle¡£¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡For¡¡illustrative¡¡purposes£»¡¡let¡¯s¡¡try¡¡a¡¡first¡¡stab¡¡at¡¡the¡¡search¡¡algorithm¡¡and¡¡see¡¡where¡¡things¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡take¡¡us¡£¡¡We¡¡could¡¡start¡¡by¡¡defining¡¡the¡¡search¡¡class¡¡or¡¡start¡¡by¡¡defining¡¡the¡¡test¡¡that¡¡will¡¡test¡¡the¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡search¡¡class¡£¡¡Let¡¯s¡¡define¡¡the¡¡test¡¡first£»¡¡because¡¡it¡¡allows¡¡us¡¡to¡¡figure¡¡out¡¡what¡¡shape¡¡the¡¡search¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡class¡¡should¡¡take£º¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Page¡¡121¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡CH¡¡AP¡¡T¡¡E¡¡R¡¡¡¡¡¡4¡¡¡¡¡¡¡ö¡¡¡¡¡¡¡¡L¡¡E¡¡A¡¡R¡¡N¡¡I¡¡N¡¡G¡¡¡¡¡¡A¡¡B¡¡OU¡¡T¡¡¡¡¡¡D¡¡AT¡¡A¡¡¡¡S¡¡TR¡¡U¡¡CT¡¡U¡¡R¡¡E¡¡S£»¡¡¡¡¡¡DE¡¡CI¡¡SI¡¡ON¡¡S£»¡¡¡¡¡¡A¡¡N¡¡D¡¡¡¡¡¡L¡¡O¡¡OP¡¡S¡¡99¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡SearchAlgorithm¡£DepthFirstFindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡In¡¡the¡¡test£»¡¡the¡¡search¡¡algorithm¡¡is¡¡called¡¡directly¡¡using¡¡¡¡
SearchAlgorithm¡£DepthFirstFindRoute£¨£©¡£¡¡Here£»¡¡SearchAlgorithm¡¡is¡¡the¡¡name¡¡of¡¡the¡¡class£»¡¡¡¡
and¡¡DepthFirstFindRoute£¨£©¡¡is¡¡the¡¡name¡¡of¡¡the¡¡method¡£¡¡The¡¡class¡¡name¡¡implies¡¡that¡¡this¡¡class¡¡¡¡
will¡¡contain¡¡all¡¡search¡¡algorithm¡¡implementations¡£¡¡This¡¡is¡¡wrong£»¡¡because¡¡most¡¡likely£»¡¡each¡¡¡¡
search¡¡algorithm¡¡will¡¡require¡¡multiple¡¡methods£»¡¡leading¡¡to¡¡a¡¡very¡¡large¡¡and¡¡plicated¡¡¡¡
SearchAlgorithm¡¡class¡£¡¡If¡¡this¡¡is¡¡the¡¡case£»¡¡then¡¡maintaining¡¡the¡¡¡¡SearchAlgorithm¡¡class¡¡will¡¡¡¡
bee¡¡a¡¡nightmare¡£¡¡
¡¡¡¡¡¡¡¡¡¡A¡¡better¡¡solution¡¡would¡¡be¡¡to¡¡identify¡¡a¡¡single¡¡class¡¡as¡¡being¡¡a¡¡single¡¡implementation¡¡of¡¡a¡¡¡¡
search¡¡algorithm¡£¡¡Then¡¡for¡¡each¡¡class£»¡¡we¡¡can¡¡define¡¡a¡¡mon¡¡method¡¡identifier¡¡that¡¡is¡¡used¡¡¡¡
to¡¡find¡¡the¡¡route¡¡between¡¡two¡¡points¡£¡¡Doing¡¡this¡¡results¡¡in¡¡the¡¡following¡¡modified¡¡test£º¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡DepthFirstSearch¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
End¡¡Sub¡¡¡¡
¡¡¡¡¡¡¡¡¡¡Now¡¡the¡¡test¡¡implies¡¡that¡¡the¡¡class¡¡DepthFirstSearch¡¡has¡¡a¡¡shared¡¡method¡¡FindRoute£¨£©¡£¡¡¡¡
This¡¡is¡¡acceptable£»¡¡and¡¡if¡¡you¡¡were¡¡to¡¡implement¡¡¡¡BreadthFirstSearch£»¡¡the¡¡naming¡¡would¡¡be¡¡¡¡
BreadthFirstSearch¡£FindRoute£¨£©¡£¡¡However£»¡¡there¡¡is¡¡another¡¡problem£»¡¡which¡¡relates¡¡to¡¡multiple¡¡¡¡
users¡¡being¡¡able¡¡to¡¡use¡¡the¡¡algorithm¡¡during¡¡the¡¡execution¡¡of¡¡a¡¡program¡£¡¡Going¡¡back¡¡to¡¡the¡¡¡¡
push¡to¡talk¡¡feature¡¡of¡¡a¡¡cell¡¡phone£»¡¡the¡¡method¡¡¡¡FindRoute£¨£©¡¡is¡¡shared¡¡and¡¡thus¡¡a¡¡shared¡¡¡¡
resource¡£¡¡If¡¡multiple¡¡users¡¡do¡¡use¡¡this¡¡algorithm£»¡¡they¡¡will¡¡share¡¡the¡¡resource¡£¡¡This¡¡could¡¡be¡¡¡¡
problematic¡¡if¡¡you¡¡are¡¡storing¡¡temporary¡¡data¡¡in¡¡the¡¡data¡¡members¡¡of¡¡the¡¡DepthFirstSearch¡¡¡¡
class¡£¡¡Using¡¡a¡¡shared¡¡method¡¡could¡¡corrupt¡¡your¡¡found¡¡search¡¡path¡£¡¡
¡¡¡¡¡¡¡¡¡¡The¡¡more¡¡appropriate¡¡solution¡¡is¡¡to¡¡define¡¡the¡¡method¡¡¡¡FindRoute£¨£©¡¡as¡¡a¡¡non¡shared¡¡¡¡
method£»¡¡implying¡¡that¡¡DepthFirstSearch¡¡must¡¡be¡¡instantiated¡¡before¡¡we¡¡can¡¡call¡¡¡¡FindRoute£¨£©¡£¡¡¡¡
We¡¡should¡¡modify¡¡the¡¡test¡¡again¡¡as¡¡follows£º¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨£©¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡To¡¡execute¡¡the¡¡method¡¡FindRoute£¨£©£»¡¡we¡¡need¡¡to¡¡create¡¡a¡¡DepthFirstSearch¡¡object£»¡¡allowing¡¡¡¡
multiple¡¡users¡¡to¡¡perform¡¡searches¡¡without¡¡getting¡¡state¡¡mixed¡¡up¡£¡¡At¡¡this¡¡point£»¡¡we¡¡could¡¡pat¡¡¡¡
ourselves¡¡on¡¡the¡¡back¡¡and¡¡think¡¡that¡¡we¡¡have¡¡written¡¡a¡¡good¡¡test¡¡that¡¡requires¡¡a¡¡class¡¡¡¡
implementation¡£¡¡
The¡¡Problem¡¡of¡¡Magic¡¡Data¡¡
Our¡¡test¡¡is¡¡not¡¡yet¡¡plete£»¡¡because¡¡we¡¡don¡¯t¡¡have¡¡access¡¡to¡¡the¡¡route¡¡found¡¡by¡¡the¡¡algorithm£»¡¡¡¡
but¡¡that¡¡will¡¡be¡¡explained¡¡in¡¡a¡¡moment¡£¡¡¡¡
¡¡¡¡¡¡¡¡¡¡In¡¡the¡¡implementation¡¡of¡¡DepthFirstSearch£»¡¡a¡¡reference¡¡to¡¡the¡¡data¡¡structure¡¡is¡¡necessary¡£¡¡¡¡
The¡¡search¡¡algorithm¡¡needs¡¡to¡¡know¡¡which¡¡tree¡¡to¡¡navigate¡£¡¡One¡¡way¡¡to¡¡implement¡¡a¡¡reference¡¡¡¡
to¡¡the¡¡tree¡¡is¡¡to¡¡directly¡¡reference¡¡the¡¡shared¡¡data¡¡Node¡£RootNodes¡£¡¡An¡¡implementation¡¡of¡¡¡¡
DepthFirstSearch¡¡would¡¡be¡¡as¡¡follows£º¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Page¡¡122¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
100¡¡¡¡¡¡¡¡¡¡¡¡¡¡CH¡¡AP¡¡T¡¡E¡¡R¡¡¡¡¡¡4¡¡¡¡¡¡¡ö¡¡¡¡¡¡¡¡L¡¡E¡¡A¡¡R¡¡N¡¡IN¡¡G¡¡¡¡¡¡AB¡¡OU¡¡T¡¡¡¡¡¡D¡¡AT¡¡A¡¡¡¡S¡¡TR¡¡U¡¡CT¡¡U¡¡R¡¡E¡¡S£»¡¡¡¡¡¡DE¡¡CI¡¡SI¡¡ON¡¡S£»¡¡¡¡¡¡A¡¡N¡¡D¡¡¡¡¡¡L¡¡O¡¡OP¡¡S¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Public¡¡Class¡¡DepthFirstSearch¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Public¡¡Sub¡¡FindRoute£¨ByVal¡¡start¡¡As¡¡String£»¡¡ByVal¡¡finish¡¡As¡¡String£©¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Dim¡¡startNodes¡¡As¡¡Node£¨£©¡¡=¡¡Node¡£RootNodes¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡End¡¡Class¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡This¡¡example¡¡declares¡¡a¡¡variable¡¡called¡¡startNodes£»¡¡which¡¡represents¡¡the¡¡starting¡¡point¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡and¡¡root¡¡of¡¡the¡¡tree¡¡as¡¡shown¡¡in¡¡Figure¡¡4¡2¡£¡¡The¡¡root¡¡of¡¡the¡¡tree¡¡is¡¡based¡¡on¡¡the¡¡data¡¡member¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Node¡£RootNodes£»¡¡and¡¡this¡¡assignment¡¡is¡¡called¡¡a¡¡magic¡¡type¡¡assignment¡£¡¡A¡¡magic¡¡type¡¡is¡¡formed¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡when¡¡you¡¡call¡¡a¡¡method£»¡¡and¡¡magically£»¡¡it¡¡happens¡¡to¡¡know¡¡how¡¡to¡¡reference¡¡data£»¡¡even¡¡though¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡you¡¡never¡¡instructed¡¡the¡¡type¡£¡¡In¡¡the¡¡case¡¡of¡¡DepthFirstSearch£»¡¡the¡¡magic¡¡is¡¡the¡¡ability¡¡of¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡FindRoute£¨£©¡¡to¡¡know¡¡to¡¡reference¡¡the¡¡correct¡¡data¡¡member¡¡RootNodes¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡The¡¡assumption¡¡is¡¡bad¡¡because¡¡it¡¡couples¡¡the¡¡data¡¡member¡¡¡¡RootNodes¡¡to¡¡the¡¡method¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡FindRoute£¨£©¡£¡¡Imagine¡¡if¡¡the¡¡developer¡¡of¡¡the¡¡Node¡¡class¡¡later¡¡decides¡¡to¡¡add¡¡functionality¡¡to¡¡load¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡the¡¡tree¡¡from¡¡a¡¡file¡¡on¡¡the¡¡hard¡¡disk¡£¡¡So¡¡that¡¡¡¡FindRoute£¨£©¡¡is¡¡not¡¡broken£»¡¡the¡¡developer¡¡would¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡need¡¡to¡¡explicitly¡¡copy¡¡the¡¡hard¡disk¡loaded¡¡tree¡¡to¡¡the¡¡data¡¡member¡¡RootNodes¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Or¡¡what¡¡if¡¡two¡¡different¡¡users¡¡wanted¡¡to¡¡create¡¡two¡¡different¡¡flight¡¡trees£¿¡¡Node¡£RootNodes¡¡is¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡a¡¡shared¡¡resource£»¡¡and¡¡thus¡¡can¡¡process¡¡only¡¡a¡¡single¡¡flight¡¡tree¡£¡¡The¡¡developer¡¡of¡¡Node¡¡might¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡alter¡¡RootNodes£»¡¡and¡¡thus¡¡FindRoute£¨£©¡¡would¡¡behave¡¡erratically¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡When¡¡you¡¡have¡¡a¡¡case¡¡of¡¡magic¡¡data£»¡¡whatever¡¡data¡¡is¡¡magic¡¡needs¡¡to¡¡be¡¡passed¡¡to¡¡the¡¡type¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡via¡¡a¡¡constructor¡¡or¡¡other¡¡method¡£¡¡So¡¡the¡¡test¡¡for¡¡the¡¡flight¡¡route¡¡would¡¡change¡¡to¡¡the¡¡following£º¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch=¡¡_¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡As¡¡the¡¡root¡¡tree¡¡node¡¡is¡¡required£»¡¡we¡¡change¡¡the¡¡constructor¡¡to¡¡require¡¡that¡¡a¡¡caller¡¡pass¡¡in¡¡the¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡root¡¡tree¡¡node¡£¡¡The¡¡test¡¡code¡¡still¡¡uses¡¡the¡¡shared¡¡data¡¡member¡¡RootNodes£»¡¡but¡¡DepthFirstSearch¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡does¡¡not¡¡need¡¡to¡¡know¡¡where¡¡to¡¡find¡¡the¡¡tree¡£¡¡If¡¡the¡¡Node¡¡developer¡¡were¡¡to¡¡alter¡¡the¡¡behavior¡¡of¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡the¡¡data¡¡member¡¡RootNodes£»¡¡then¡¡only¡¡the¡¡constructor¡¡code¡¡to¡¡DepthFirstSearch¡¡would¡¡need¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡altering£»¡¡not¡¡the¡¡FindRoute£¨£©¡¡method¡£¡¡Thus£»¡¡Node¡¡and¡¡DepthFirstSearch¡¡are¡¡properly¡¡decoupled¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡from¡¡each¡¡other¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Getting¡¡the¡¡Found¡¡Route¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Once¡¡you¡¡have¡¡called¡¡the¡¡FindRoute£¨£©¡¡method£»¡¡you¡¡expect¡¡an¡¡answer¡£¡¡Because¡¡the¡¡route¡¡could¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡involve¡¡multiple¡¡cities£»¡¡the¡¡found¡¡route¡¡is¡¡stored¡¡in¡¡an¡¡array¡¡of¡¡Node¡¡elements¡£¡¡In¡¡programmatic¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡terms£»¡¡there¡¡are¡¡two¡¡ways¡¡of¡¡retrieving¡¡the¡¡array¡¡of¡¡Nodes¡£¡¡The¡¡first¡¡is¡¡a¡¡return¡¡value£»¡¡like¡¡this£º¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Dim¡¡foundRoute¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡The¡¡bold¡¡code¡¡shows¡¡the¡¡assignment¡¡of¡¡the¡¡return¡¡value¡¡to¡¡the¡¡variable¡¡foundRoute¡£¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Page¡¡123¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡CH¡¡AP¡¡T¡¡E¡¡R¡¡¡¡¡¡4¡¡¡¡¡¡¡ö¡¡¡¡¡¡¡¡L¡¡E¡¡A¡¡R¡¡N¡¡I¡¡N¡¡G¡¡¡¡¡¡A¡¡B¡¡OU¡¡T¡¡¡¡¡¡D¡¡AT¡¡A¡¡¡¡S¡¡TR¡¡U¡¡CT¡¡U¡¡R¡¡E¡¡S£»¡¡¡¡¡¡DE¡¡CI¡¡SI¡¡ON¡¡S£»¡¡¡¡¡¡A¡¡N¡¡D¡¡¡¡¡¡L¡¡O¡¡OP¡¡S¡¡101¡¡
¡¡¡¡¡¡¡¡¡¡The¡¡second¡¡approach¡¡is¡¡for¡¡the¡¡FindRoute£¨£©¡¡method¡¡to¡¡store¡¡the¡¡result¡¡in¡¡an¡¡internal¡¡data¡¡¡¡
member¡£¡¡The¡¡following¡¡example¡¡assumes¡¡that¡¡the¡¡FindRoute£¨£©¡¡method¡¡stores¡¡the¡¡result¡¡in¡¡a¡¡¡¡
data¡¡member¡¡named¡¡FoundRoute£º¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡Dim¡¡cls¡¡As¡¡DepthfirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡Dim¡¡foundRoute¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FoundRoute¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡Each¡¡approach¡¡seems¡¡acceptable£»¡¡and¡¡you¡¡are¡¡not¡¡sure¡¡which¡¡to¡¡use¡£¡¡When¡¡you¡¡have¡¡a¡¡¡¡
choice¡¡like¡¡this£»¡¡you¡¡need¡¡to¡¡make¡¡a¡¡decision¡£¡¡The¡¡safest¡¡way¡¡to¡¡make¡¡a¡¡decision¡¡is¡¡to¡¡write¡¡tests¡¡¡¡
and¡¡see¡¡if¡¡there¡¡are¡¡any¡¡problems¡¡with¡¡either¡¡approach¡£¡¡
¡¡¡¡¡¡¡¡¡¡In¡¡the¡¡example¡¡of¡¡calculating¡¡a¡¡single¡¡route£»¡¡either¡¡approach¡¡is¡¡fine¡£¡¡But¡¡let¡¯s¡¡look¡¡at¡¡the¡¡¡¡
code¡¡when¡¡multiple¡¡routes¡¡are¡¡being¡¡searched¡£¡¡First£»¡¡consider¡¡the¡¡code¡¡where¡¡the¡¡found¡¡path¡¡¡¡
is¡¡a¡¡return¡¡parameter¡¡value£º¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡Dim¡¡foundRoute1¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡Dim¡¡foundRoute2¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FindRoute£¨¡¨New¡¡York¡¨£»¡¡¡¨Seattle¡¨£©¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡Now¡¡take¡¡a¡¡look¡¡at¡¡the¡¡code¡¡that¡¡uses¡¡the¡¡data¡¡member£º¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡Dim¡¡foundRoute1¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FoundRoute¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨New¡¡York¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡Dim¡¡foundRoute2¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FoundRoute¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡Again£»¡¡it¡¡would¡¡seem¡¡that¡¡both¡¡choices¡¡are¡¡adequate¡£¡¡However£»¡¡there¡¡is¡¡a¡¡difference£»¡¡the¡¡¡¡
difference¡¡is¡¡subtle£»¡¡but¡¡distinct¡¡enough¡¡to¡¡matter¡£¡¡In¡¡the¡¡test¡¡implementation¡¡where¡¡the¡¡found¡¡¡¡
route¡¡is¡¡a¡¡return¡¡value£»¡¡the¡¡variables¡¡foundRoute1¡¡and¡¡foundRoute2¡¡represent¡¡routes¡¡that¡¡relate¡¡¡¡
directly¡¡to¡¡the¡¡route¡¡being¡¡searched¡£¡¡There¡¡is¡¡no¡¡chance¡¡that¡¡the¡¡variables¡¡foundRoute1¡¡can¡¡¡¡
represent¡¡the¡¡route¡¡New¡¡York¨CSeattle¡£¡¡With¡¡the¡¡data¡¡member¡¡code£»¡¡it¡¡could¡¡happen¡¡that¡¡¡¡
foundRoute1¡¡points¡¡to¡¡the¡¡route¡¡New¡¡York¨CSeattle£»¡¡as¡¡shown¡¡in¡¡the¡¡following¡¡code¡£¡¡
Public¡¡Sub¡¡TestSearch£¨£©¡¡¡¡
¡¡¡¡Dim¡¡cls¡¡As¡¡DepthFirstSearch¡¡=¡¡New¡¡DepthFirstSearch£¨Node¡£RootNodes£©¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨Montreal¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡cls¡£FindRoute£¨¡¨New¡¡York¡¨£»¡¡¡¨Seattle¡¨£©¡¡
¡¡¡¡Dim¡¡foundRoute1¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FoundRoute¡¡
¡¡¡¡Dim¡¡foundRoute2¡¡As¡¡Node£¨£©¡¡=¡¡cls¡£FoundRoute¡¡
End¡¡Sub¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Page¡¡124¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
102¡¡¡¡¡¡¡¡¡¡¡¡¡¡CH¡¡AP¡¡T¡¡E¡¡R¡¡¡¡¡¡4¡¡¡¡¡¡¡ö¡¡¡¡¡¡¡¡L¡¡E¡¡A¡¡R¡¡N¡¡IN¡¡G¡¡¡¡¡¡AB¡¡OU¡¡T¡¡¡¡¡¡D¡¡AT¡¡A¡¡¡¡S¡¡TR¡¡U¡¡CT¡¡U¡¡R¡¡E¡¡S£»¡¡¡¡¡¡DE¡¡CI¡¡SI¡¡ON¡¡S£»¡¡¡¡¡¡A¡¡N¡¡D¡¡¡¡¡¡L¡¡O¡¡OP¡¡S¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡By¡¡switching¡¡the¡¡order¡¡of¡¡the¡¡FindRoute£¨£©¡¡method¡¡calls¡¡and¡¡references¡¡to¡¡the¡¡data¡¡member¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡FoundRoute£»¡¡the¡¡variables¡¡foundRoute1¡¡and¡¡foundRoute2¡¡will¡¡reference¡¡the¡¡same¡¡found¡¡route£»¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡specifically¡¡the¡¡route¡¡New¡¡York¨CSeattle¡£¡¡This¡¡is¡¡not¡¡a¡¡good¡¡idea¡£¡¡The¡¡example¡¡shows¡¡how¡¡data¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡members¡¡have¡¡no¡¡direct¡¡relation¡¡to¡¡methods¡¡and¡¡can¡¡vary¡¡independently¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡So¡¡the¡¡choice¡¡of¡¡returning¡¡the¡¡found¡¡route¡¡from¡¡a¡¡method¡¡is¡¡the¡¡better¡¡and¡¡more¡¡robust¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡approach¡£¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡öNote¡¡¡¡Data¡¡members¡¡are¡¡useful¡¡when¡¡you¡¡want¡¡to¡¡store¡¡or¡¡retrieve¡¡data¡¡that¡¡spans¡¡multiple¡¡method¡¡calls¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡or¡¡is¡¡not¡¡dependent¡¡on¡¡the¡¡order¡¡of¡¡how¡¡methods¡¡are¡¡called¡£¡¡When¡¡you¡¡have¡¡data¡¡that¡¡is¡¡dependent¡¡on¡¡the¡¡¡¡
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡order¡¡of¡¡c
¿ì½Ý²Ù×÷: °´¼üÅÌÉÏ·½Ïò¼ü ¡û »ò ¡ú ¿É¿ìËÙÉÏÏ·ҳ °´¼üÅÌÉ쵀 Enter ¼ü¿É»Øµ½±¾ÊéĿ¼ҳ °´¼üÅÌÉÏ·½Ïò¼ü ¡ü ¿É»Øµ½±¾Ò³¶¥²¿!
ÎÂÜ°Ìáʾ£º ο´Ð¡ËµµÄͬʱ·¢±íÆÀÂÛ£¬Ëµ³ö×Ô¼ºµÄ¿´·¨ºÍÆäËüС»ï°éÃÇ·ÖÏíÒ²²»´íŶ£¡·¢±íÊéÆÀ»¹¿ÉÒÔ»ñµÃ»ý·ÖºÍ¾Ñé½±Àø£¬ÈÏÕæдԴ´ÊéÆÀ ±»²ÉÄÉΪ¾«ÆÀ¿ÉÒÔ»ñµÃ´óÁ¿½ð±Ò¡¢»ý·ÖºÍ¾Ñé½±ÀøŶ£¡