Merge joreland@bk-internal.mysql.com:/home/bk/mysql-4.1

into  perch.ndb.mysql.com:/home/jonas/src/mysql-4.1-push

5 files changed, 150 insertions, 19 deletions

diff --git a/ndb/docs/Makefile.am b/ndb/docs/Makefile.am
index 1399ce3b6a5..9b7ec213e26 100644
--- a/ndb/docs/Makefile.am
+++ b/ndb/docs/Makefile.am
@@ -1,14 +1,12 @@
 DOXYDIR = doxygen
 noinst_HEADERS = $(DOXYDIR)/predoxy.pl $(DOXYDIR)/postdoxy.pl $(DOXYDIR)/Doxyfile.ndbapi $(DOXYDIR)/Doxyfile.mgmapi $(DOXYDIR)/header.ndbapi.tex $(DOXYDIR)/header.mgmapi.tex
 
-all: 	do-check ndbapidoc mgmapidoc
-
 DOXYTMP =	.doxytmp
 DOXYOUT =	.doxyout
 
 NDB_RELEASE = @NDB_VERSION_MAJOR@.@NDB_VERSION_MINOR@.@NDB_VERSION_BUILD@-@NDB_VERSION_STATUS@
 
-clean:
+clean-local:
 	rm -rf ndbapi.pdf ndbapi.html mgmapi.pdf mgmapi.html
 	rm -rf $(DOXYTMP) $(DOXYOUT)
 
diff --git a/ndb/include/kernel/AttributeDescriptor.hpp b/ndb/include/kernel/AttributeDescriptor.hpp
index 071d45e2607..9d7de21d904 100644
--- a/ndb/include/kernel/AttributeDescriptor.hpp
+++ b/ndb/include/kernel/AttributeDescriptor.hpp
@@ -36,6 +36,7 @@ private:
   
   static Uint32 getType(const Uint32 &);
   static Uint32 getSize(const Uint32 &);
+  static Uint32 getSizeInBytes(const Uint32 &);
   static Uint32 getSizeInWords(const Uint32 &);
   static Uint32 getArrayType(const Uint32 &);
   static Uint32 getArraySize(const Uint32 &);
@@ -79,6 +80,7 @@ private:
 #define AD_SIZE_SHIFT       (4)
 #define AD_SIZE_MASK        (7)
 
+#define AD_SIZE_IN_BYTES_SHIFT (3)
 #define AD_SIZE_IN_WORDS_OFFSET (31)
 #define AD_SIZE_IN_WORDS_SHIFT  (5)
 
@@ -187,6 +189,13 @@ AttributeDescriptor::getSize(const Uint32 & desc){
 
 inline
 Uint32
+AttributeDescriptor::getSizeInBytes(const Uint32 & desc){
+  return (getArraySize(desc) << getSize(desc)) 
+                             >> AD_SIZE_IN_BYTES_SHIFT;
+}
+
+inline
+Uint32
 AttributeDescriptor::getSizeInWords(const Uint32 & desc){
   return ((getArraySize(desc) << getSize(desc)) 
           + AD_SIZE_IN_WORDS_OFFSET) 
diff --git a/ndb/src/kernel/blocks/dbtup/DbtupRoutines.cpp b/ndb/src/kernel/blocks/dbtup/DbtupRoutines.cpp
index cbb165c3eb1..7b642f90a17 100644
--- a/ndb/src/kernel/blocks/dbtup/DbtupRoutines.cpp
+++ b/ndb/src/kernel/blocks/dbtup/DbtupRoutines.cpp
@@ -700,6 +700,27 @@ Dbtup::checkUpdateOfPrimaryKey(Uint32* updateBuffer, Tablerec* const regTabPtr)
   Uint32 attrDescriptorIndex = regTabPtr->tabDescriptor + (attributeId << ZAD_LOG_SIZE);
   Uint32 attrDescriptor = tableDescriptor[attrDescriptorIndex].tabDescr;
   Uint32 attributeOffset = tableDescriptor[attrDescriptorIndex + 1].tabDescr;
+
+  Uint32 xfrmBuffer[1 + MAX_KEY_SIZE_IN_WORDS * 1]; // strxfrm_multiply == 1
+  Uint32 charsetFlag = AttributeOffset::getCharsetFlag(attributeOffset);
+  if (charsetFlag) {
+    Uint32 csPos = AttributeOffset::getCharsetPos(attributeOffset);
+    CHARSET_INFO* cs = regTabPtr->charsetArray[csPos];
+    Uint32 sizeInBytes = AttributeDescriptor::getSizeInBytes(attrDescriptor);
+    Uint32 sizeInWords = AttributeDescriptor::getSizeInWords(attrDescriptor);
+    const uchar* srcPtr = (uchar*)&updateBuffer[1];
+    uchar* dstPtr = (uchar*)&xfrmBuffer[1];
+    Uint32 n =
+      (*cs->coll->strnxfrm)(cs, dstPtr, sizeInBytes, srcPtr, sizeInBytes);
+    // pad with blanks (unlikely) and zeroes to match NDB API behaviour
+    while (n < sizeInBytes)
+      dstPtr[n++] = 0x20;
+    while (n < 4 * sizeInWords)
+      dstPtr[n++] = 0;
+    xfrmBuffer[0] = ahIn.m_value;
+    updateBuffer = xfrmBuffer;
+  }
+
   ReadFunction f = regTabPtr->readFunctionArray[attributeId];
 
   AttributeHeader::init(&attributeHeader, attributeId, 0);
@@ -707,7 +728,7 @@ Dbtup::checkUpdateOfPrimaryKey(Uint32* updateBuffer, Tablerec* const regTabPtr)
   tMaxRead = MAX_KEY_SIZE_IN_WORDS;
 
   bool tmp = tXfrmFlag;
-  tXfrmFlag = false;
+  tXfrmFlag = true;
   ndbrequire((this->*f)(&keyReadBuffer[0], ahOut, attrDescriptor, attributeOffset));
   tXfrmFlag = tmp;
   ndbrequire(tOutBufIndex == ahOut->getDataSize());
diff --git a/ndb/tools/ndb_size.pl b/ndb/tools/ndb_size.pl
index 64a20423636..ece0901e0b2 100644
--- a/ndb/tools/ndb_size.pl
+++ b/ndb/tools/ndb_size.pl
@@ -26,24 +26,33 @@ use HTML::Template;
 # BUGS
 # ----
 # - enum/set is 0 byte storage! Woah - efficient!
+# - DECIMAL is 0 byte storage. A bit too efficient.
 # - some float stores come out weird (when there's a comma e.g. 'float(4,1)')
 # - no disk data values
 # - computes the storage requirements of views (and probably MERGE)
 # - ignores character sets.
 
 my $template = HTML::Template->new(filename => 'ndb_size.tmpl',
-				   die_on_bad_params => 0);
+				   die_on_bad_params => 0)
+    or die "Could not open ndb_size.tmpl.";
 
 my $dbh;
 
+if(@ARGV < 3 || $ARGV[0] eq '--usage' || $ARGV[0] eq '--help')
+{
+    print STDERR "Usage:\n";
+    print STDERR "\tndb_size.pl database hostname user password\n\n";
+    print STDERR "If you need to specify a port number, use host:port\n\n";
+    exit(1);
+}
+
 {
     my $database= $ARGV[0];
     my $hostname= $ARGV[1];
-    my $port= $ARGV[2];
-    my $user= $ARGV[3];
-    my $password= $ARGV[4];
-    my $dsn = "DBI:mysql:database=$database;host=$hostname;port=$port";
-    $dbh= DBI->connect($dsn, $user, $password);
+    my $user= $ARGV[2];
+    my $password= $ARGV[3];
+    my $dsn = "DBI:mysql:database=$database;host=$hostname";
+    $dbh= DBI->connect($dsn, $user, $password) or exit(1);
     $template->param(db => $database);
     $template->param(dsn => $dsn);
 }
@@ -55,6 +64,14 @@ my $tables  = $dbh->selectall_arrayref("show tables");
 
 my @table_size;
 
+my @dbDataMemory;
+my @dbIndexMemory;
+my @NoOfAttributes;
+my @NoOfIndexes;
+my @NoOfTables;
+$NoOfTables[$_]{val} = @{$tables} foreach 0..$#releases;
+
+
 sub align {
     my($to,@unaligned) = @_;
     my @aligned;
@@ -68,9 +85,8 @@ foreach(@{$tables})
 {
     my $table= @{$_}[0];
     my @columns;
-    my $info= $dbh->selectall_hashref("describe ".$dbh->quote($table),"Field");
-    my @count  = $dbh->selectrow_array("select count(*) from "
-				       .$dbh->quote($table));
+    my $info= $dbh->selectall_hashref('describe `'.$table.'`',"Field");
+    my @count  = $dbh->selectrow_array('select count(*) from `'.$table.'`');
     my %columnsize; # used for index calculations
 
     # We now work out the DataMemory usage
@@ -132,14 +148,17 @@ foreach(@{$tables})
 	    my $fixed= 1+$size;
 	    my @dynamic=$dbh->selectrow_array("select avg(length("
 					      .$dbh->quote($name)
-					      .")) from ".$dbh->quote($table));
+					      .")) from `".$table.'`');
 	    $dynamic[0]=0 if !$dynamic[0];
 	    @realsize= ($fixed,$fixed,ceil($dynamic[0]));
 	}
 	elsif($type =~ /binary/ || $type =~ /char/)
 	{@realsize=($size,$size,$size)}
 	elsif($type =~ /text/ || $type =~ /blob/)
-	{@realsize=(256,256,1)} # FIXME check if 5.1 is correct
+	{
+	    @realsize=(256,256,1);
+	    $NoOfTables[$_]{val} += 1 foreach 0..$#releases; # blob uses table
+	} # FIXME check if 5.1 is correct
 
 	@realsize= align(4,@realsize);
 
@@ -166,7 +185,7 @@ foreach(@{$tables})
     # we can still connect to pre-5.0 mysqlds.
     my %indexes;
     {
-	my $sth= $dbh->prepare("show index from "$dbh->quote($table));
+	my $sth= $dbh->prepare("show index from `".$table.'`');
 	$sth->execute;
 	while(my $i = $sth->fetchrow_hashref)
 	{    
@@ -257,7 +276,51 @@ foreach(@{$tables})
 	IndexMemory=>\@IndexMemory,
 	
     };
+
+    $dbDataMemory[$_]{val} += $DataMemory[$_]{val} foreach 0..$#releases;
+    $dbIndexMemory[$_]{val} += $IndexMemory[$_]{val} foreach 0..$#releases;
+    $NoOfAttributes[$_]{val} += @columns foreach 0..$#releases;
+    $NoOfIndexes[$_]{val} += @indexes foreach 0..$#releases;
+}
+
+my @NoOfTriggers;
+# for unique hash indexes
+$NoOfTriggers[$_]{val} += $NoOfIndexes[$_]{val}*3 foreach 0..$#releases;
+# for ordered index
+$NoOfTriggers[$_]{val} += $NoOfIndexes[$_]{val} foreach 0..$#releases;
+
+my @ParamMemory;
+foreach (0..$#releases) {
+    $ParamMemory[0]{releases}[$_]{val}= POSIX::ceil(200*$NoOfAttributes[$_]{val}/1024);
+    $ParamMemory[0]{name}= 'Attributes';
+
+    $ParamMemory[1]{releases}[$_]{val}= 20*$NoOfTables[$_]{val};
+    $ParamMemory[1]{name}= 'Tables';
+
+    $ParamMemory[2]{releases}[$_]{val}= 10*$NoOfIndexes[$_]{val};
+    $ParamMemory[2]{name}= 'OrderedIndexes';
+
+    $ParamMemory[3]{releases}[$_]{val}= 15*$NoOfIndexes[$_]{val};
+    $ParamMemory[3]{name}= 'UniqueHashIndexes';    
 }
 
 $template->param(tables => \@table_size);
+$template->param(Parameters => [{name=>'DataMemory (kb)',
+				 releases=>\@dbDataMemory},
+				{name=>'IndexMemory (kb)',
+				 releases=>\@dbIndexMemory},
+				{name=>'MaxNoOfTables',
+				 releases=>\@NoOfTables},
+				{name=>'MaxNoOfAttributes',
+				 releases=>\@NoOfAttributes},
+				{name=>'MaxNoOfOrderedIndexes',
+				 releases=>\@NoOfIndexes},
+				{name=>'MaxNoOfUniqueHashIndexes',
+				 releases=>\@NoOfIndexes},
+				{name=>'MaxNoOfTriggers',
+				 releases=>\@NoOfTriggers}
+				]
+		 );
+$template->param(ParamMemory => \@ParamMemory);
+
 print $template->output;
diff --git a/ndb/tools/ndb_size.tmpl b/ndb/tools/ndb_size.tmpl
index d83d5d2c6af..dc02b5a5970 100644
--- a/ndb/tools/ndb_size.tmpl
+++ b/ndb/tools/ndb_size.tmpl
@@ -13,18 +13,58 @@ td,th { border: 1px solid black }
 <h1>MySQL Cluster analysis for <TMPL_VAR NAME="db" escape="html"></h1>
 <p>This is an automated analysis of the <TMPL_VAR NAME="DSN" escape="html"> database for migration into <a href="http://www.mysql.com/">MySQL</a> Cluster. No warranty is made to the accuracy of the information.</p>
 
-<p>This information should be valid for MySQL 4.1</p>
+<p>This information should be valid for MySQL 4.1 and 5.0. Since 5.1 is not a final release yet, the numbers should be used as a guide only.</p>
 
+<h2>Parameter Settings</h2>
+<p><b>NOTE</b> the configuration parameters below do not take into account system tables and other requirements.</p>
+<table>
+  <tr>
+    <th>Parameter</th>
+    <TMPL_LOOP NAME=releases>
+      <th><TMPL_VAR NAME=rel></th>
+    </TMPL_LOOP>
+  </tr>
+<TMPL_LOOP NAME=Parameters>
+  <tr>
+    <td><TMPL_VAR NAME=name></td>
+    <TMPL_LOOP NAME=releases>
+      <td><TMPL_VAR NAME=val></td>
+    </TMPL_LOOP>
+  </tr>
+</TMPL_LOOP>
+</table>
+
+<h2>Memory usage because of parameters</h2>
+
+<p>Usage is in kilobytes. Actual usage will vary as you should set the parameters larger than those listed in the table above.</p>
+<table>
+  <tr>
+    <th>Parameter</th>
+    <TMPL_LOOP NAME=releases>
+      <th><TMPL_VAR NAME=rel></th>
+    </TMPL_LOOP>
+  </tr>
+<TMPL_LOOP NAME=ParamMemory>
+  <tr>
+    <td><TMPL_VAR NAME=name></td>
+    <TMPL_LOOP NAME=releases>
+      <td><TMPL_VAR NAME=val></td>
+    </TMPL_LOOP>
+  </tr>
+</TMPL_LOOP>
+</table>
+
+<h2>Table List</h2>
 <ul>
 <TMPL_LOOP NAME="tables">
-<li><TMPL_VAR NAME="table"></li>
+<li><a href="#<TMPL_VAR NAME="table">"><TMPL_VAR NAME="table"></a></li>
 </TMPL_LOOP>
 </ul>
 
 <hr/>
 
 <TMPL_LOOP NAME="tables">
-<h2><TMPL_VAR NAME="table"></h2>
+<h2><a name="<TMPL_VAR NAME="table">"><TMPL_VAR NAME="table"></a></h2>
 <table>
   <tr>
     <th>Column</th>