Comments on: Encyclopedia Spine Problem http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/ Jeremy Schneider Wed, 30 Nov 2011 21:26:13 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Jeremy http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-214 Jeremy Tue, 22 May 2007 20:58:14 +0000 http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-214 Sheesh... how did I miss the ntile() function. Duuh. Just noticed that everyone else used that, so I updated my query to do the same. Also updated a few variable names to make it easier to follow. And realized that I wasn't even using the row_number(), so I removed it. Somehow problems like this are just never "done" for me. :) Sheesh… how did I miss the ntile() function. Duuh. Just noticed that everyone else used that, so I updated my query to do the same. Also updated a few variable names to make it easier to follow. And realized that I wasn’t even using the row_number(), so I removed it. Somehow problems like this are just never “done” for me. :)

]]> By: Dominic Delmolino http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-213 Dominic Delmolino Tue, 22 May 2007 19:01:25 +0000 http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-213 Pretty cool. Yeah, it's easy once you read and understand that paper. But I'm not sure everyone speaks Regex so well... Pretty cool. Yeah, it’s easy once you read and understand that paper. But I’m not sure everyone speaks Regex so well…

]]> By: Jeremy http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-208 Jeremy Tue, 22 May 2007 12:52:15 +0000 http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-208 Oops, bitten by the cut-and-paste-backslash-eater again. I copied the query into wordpress and it ate all my backslashes. Should be correct now, I escaped them. Your output actually makes perfect sense because this was the only line affected (there was a second but it's almost identical): <pre><code>'^'||rpad(regexp_replace(sn_prev,'(.)','(\1'),length(sn_prev)*4,')?') search_sn, </code></pre> It was missing the backslash which effectively made the search string "^(1(1(1)?)?)?" rather than "^(C(O(L)?)?)?" ... which obviously isn't going to find any matches - therefor the algorithm thought that there were no Common Leading Strings and sent back only the first character! Anyway it should work now... On a side note, I also tried using the extended regexp syntax "^((?:C(?:O(?:L)?)?)?)" as was discussed in <a href="http://people.cornell.edu/pages/tco2/papers/common_leading_substring/2000-06-09-0830" rel="nofollow">the CLS paper</a> to prevent capturing on all those internal parenthesis... but Oracle doesn't seem to support that operator for regular expressions. Oops, bitten by the cut-and-paste-backslash-eater again. I copied the query into wordpress and it ate all my backslashes. Should be correct now, I escaped them.

Your output actually makes perfect sense because this was the only line affected (there was a second but it’s almost identical):

'^'||rpad(regexp_replace(sn_prev,'(.)','(\1'),length(sn_prev)*4,')?') search_sn,

It was missing the backslash which effectively made the search string “^(1(1(1)?)?)?” rather than “^(C(O(L)?)?)?” … which obviously isn’t going to find any matches – therefor the algorithm thought that there were no Common Leading Strings and sent back only the first character!

Anyway it should work now…

On a side note, I also tried using the extended regexp syntax “^((?:C(?:O(?:L)?)?)?)” as was discussed in the CLS paper to prevent capturing on all those internal parenthesis… but Oracle doesn’t seem to support that operator for regular expressions.

]]> By: Dominic Brooks http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-207 Dominic Brooks Tue, 22 May 2007 08:26:10 +0000 http://www.ardentperf.com/2007/05/21/encyclopedia-spine-problem/#comment-207 Hi, Just cut and pasted your solution into 10.2.0.3.0 on Linux - the spines were coming out funny: SPINE,SN,EN #... thru B...,#_OF_PRODUCTS,BLOCKSZ B... thru C...,BLOCKS_CACHED,CONSTR_NAME C... thru D...,CONSUMED_CPU_TIME,DEFMAXEXTS D... thru E...,DEFMAXTRANS,ENQUEUE_TIME E... thru F...,ENQ_CSCN,FROM_VERSION F... thru I...,FR_OPERATIONS,INTALG I... thru L...,INTCOL,LINSTANCE# L... thru M...,LISTENER,MRCT_BASELINE_ID M... thru O...,MRCT_PURGE_TIME_NUM,OPEXPR_C1 O... thru P...,OPEXPR_N1,POLOWN P... thru R...,POLPKG,REF_MAKE_USER R... thru S...,REF_NAME,SERVICE_HASH S... thru S...,SERVICE_ID,STORAGE S... thru T...,STORAGESIZE,TRUE_RULES T... thru Z...,TRUNCATED,ZERO_RESULTS Hi,

Just cut and pasted your solution into 10.2.0.3.0 on Linux – the spines were coming out funny:

SPINE,SN,EN
#… thru B…,#_OF_PRODUCTS,BLOCKSZ
B… thru C…,BLOCKS_CACHED,CONSTR_NAME
C… thru D…,CONSUMED_CPU_TIME,DEFMAXEXTS
D… thru E…,DEFMAXTRANS,ENQUEUE_TIME
E… thru F…,ENQ_CSCN,FROM_VERSION
F… thru I…,FR_OPERATIONS,INTALG
I… thru L…,INTCOL,LINSTANCE#
L… thru M…,LISTENER,MRCT_BASELINE_ID
M… thru O…,MRCT_PURGE_TIME_NUM,OPEXPR_C1
O… thru P…,OPEXPR_N1,POLOWN
P… thru R…,POLPKG,REF_MAKE_USER
R… thru S…,REF_NAME,SERVICE_HASH
S… thru S…,SERVICE_ID,STORAGE
S… thru T…,STORAGESIZE,TRUE_RULES
T… thru Z…,TRUNCATED,ZERO_RESULTS

]]>