Commit 429f9d05 authored by unknown's avatar unknown

Merge bk-internal.mysql.com:/home/bk/mysql-4.0

into mysql.com:/home/my/mysql-4.0

parents 3db93f80 a2c50670
......@@ -65,6 +65,7 @@ marko@hundin.mysql.fi
miguel@hegel.(none)
miguel@hegel.br
miguel@hegel.local
miguel@hegel.txg
miguel@light.
miguel@light.local
miguel@sartre.local
......@@ -86,6 +87,7 @@ monty@tramp.mysql.fi
monty@work.mysql.com
mwagner@cash.mwagner.org
mwagner@evoq.mwagner.org
mwagner@here.mwagner.org
mwagner@work.mysql.com
mysqldev@build.mysql2.com
nick@mysql.com
......
#!/usr/bin/perl
#
# my_md5sum
#
# Script to clone the 'md5sum' command found on modern systems, since that
# command is not always found on all systems.
#
# Use the "--help" option for more info!
#
# Written by Matt Wagner <matt@mysql.com>
#
use strict;
use Digest::MD5;
use Getopt::Long;
my $VER= "1.1";
#
# Strip the leading path info off the program name ($0). We want 'my_md5sum'
# not './my_md5sum'.
#
$0=~ s/^.*\/(.+)$/$1/;
my ($opt_check, $opt_help)= undef;
GetOptions(
"check|c" => \$opt_check,
"help|h" => \$opt_help,
) || usage();
#
# Put all the [file1 file2 file3 ...]'s into an array
#
my @files = @ARGV;
#
# Give the "--help" text if:
# - "--help|-h" was specified
# - The number of files given as arguments is nil
# - The "--check|-c" option is used with more than one [file] argument
#
usage() if $opt_help || $#files == -1 || ($opt_check && $#files > 0);
# If "--check|-c", then go into checking
if ($opt_check)
{
open (CHECKFILE, $files[0]) or die "$files[0]: $!";
while (<CHECKFILE>)
{
#
# Goto the next line in the file if it does not match a typical
# digest line like:
#
# f1007efa2c72daa693981ec764cdeaca Bootstrap
#
next if $_!~ m/^([a-z0-9]{32})\s+(.+)$/;
# Collect the trappings from the above regex
my $checksum= $1;
my $checkfile= $2;
# Generate a fresh MD5 for the file in question
my $digest= &mkmd5($checkfile);
# Check the fresh MD5 against what is recorded in the file
# Print an error message if they don't match, else print OK
print "$checkfile: FAILED\n" if $digest ne $checksum;
print "$checkfile: OK\n" if $digest eq $checksum;
}
}
# Else generate the MD5 digest to STDOUT
else
{
foreach my $file (@files)
{
my $digest= &mkmd5($file);
print "$digest $file\n";
}
}
#
# This routine generates the MD5 digest of a file
#
sub mkmd5
{
my $file= shift;
open (FILE, $file) or die "$file: $!";
binmode(FILE);
my $digest= Digest::MD5->new->addfile(*FILE)->hexdigest;
close FILE;
return $digest;
}
#
# Print the help text
#
sub usage
{
print <<EOF;
$0 version $VER by Matt Wagner <matt\@mysql.com>
Usage:
$0 [-c [file]] | [file1...]
Generates or checks MD5 message digests.
Options:
-c, --check Check message digests (default is generate)
-h, --help Display this text and exit
The input for -c should be the list of message digests and file names that is
printed on STDOUT by this program when it generates digests.
EOF
exit(0);
}
......@@ -737,14 +737,15 @@ AC_DEFUN(MYSQL_FIND_OPENSSL, [
---)
for d in /usr/ssl/include /usr/local/ssl/include /usr/include \
/usr/include/ssl /opt/ssl/include /opt/openssl/include \
/usr/local/ssl/include /usr/local/include ; do
/usr/local/ssl/include /usr/local/include /usr/freeware/include ; do
if test -f $d/openssl/ssl.h ; then
OPENSSL_INCLUDE=-I$d
fi
done
for d in /usr/ssl/lib /usr/local/ssl/lib /usr/lib/openssl \
/usr/lib /usr/lib64 /opt/ssl/lib /opt/openssl/lib /usr/local/lib/ ; do
/usr/lib /usr/lib64 /opt/ssl/lib /opt/openssl/lib \
/usr/freeware/lib32 /usr/local/lib/ ; do
if test -f $d/libssl.a || test -f $d/libssl.so || test -f $d/libssl.dylib ; then
OPENSSL_LIB=$d
fi
......
......@@ -233,7 +233,7 @@ int main(int argc,char *argv[])
else
{
if (verbose)
printf("MySql error: %3d = %s\n",code,msg);
printf("MySQL error: %3d = %s\n",code,msg);
else
puts(msg);
}
......
......@@ -89,7 +89,7 @@ mach_read_from_4(
/* out: ulint integer */
byte* b); /* in: pointer to four bytes */
/*************************************************************
Writes a ulint in a compressed form. */
Writes a ulint in a compressed form (1..5 bytes). */
UNIV_INLINE
ulint
mach_write_compressed(
......@@ -168,7 +168,7 @@ mach_read_from_8(
/* out: dulint integer */
byte* b); /* in: pointer to 8 bytes */
/*************************************************************
Writes a dulint in a compressed form. */
Writes a dulint in a compressed form (5..9 bytes). */
UNIV_INLINE
ulint
mach_dulint_write_compressed(
......@@ -193,7 +193,7 @@ mach_dulint_read_compressed(
/* out: read dulint */
byte* b); /* in: pointer to memory from where to read */
/*************************************************************
Writes a dulint in a compressed form. */
Writes a dulint in a compressed form (1..11 bytes). */
UNIV_INLINE
ulint
mach_dulint_write_much_compressed(
......
......@@ -366,7 +366,7 @@ mach_read_from_6(
}
/*************************************************************
Writes a dulint in a compressed form. */
Writes a dulint in a compressed form (5..9 bytes). */
UNIV_INLINE
ulint
mach_dulint_write_compressed(
......@@ -422,7 +422,7 @@ mach_dulint_read_compressed(
}
/*************************************************************
Writes a dulint in a compressed form. */
Writes a dulint in a compressed form (1..11 bytes). */
UNIV_INLINE
ulint
mach_dulint_write_much_compressed(
......
......@@ -108,7 +108,9 @@ mlog_close(
mtr_t* mtr, /* in: mtr */
byte* ptr); /* in: buffer space from ptr up was not used */
/************************************************************
Writes the initial part of a log record. */
Writes the initial part of a log record (3..11 bytes).
If the implementation of this function is changed, all
size parameters to mlog_open() should be adjusted accordingly! */
UNIV_INLINE
byte*
mlog_write_initial_log_record_fast(
......
......@@ -137,7 +137,9 @@ mlog_catenate_dulint_compressed(
}
/************************************************************
Writes the initial part of a log record. */
Writes the initial part of a log record (3..11 bytes).
If the implementation of this function is changed, all
size parameters to mlog_open() should be adjusted accordingly! */
UNIV_INLINE
byte*
mlog_write_initial_log_record_fast(
......
......@@ -20,8 +20,9 @@ Protected by mem_hash_mutex above. */
static ulint mem_n_created_heaps = 0;
static ulint mem_n_allocations = 0;
static ulint mem_total_allocated_memory = 0;
ulint mem_current_allocated_memory = 0;
ulint mem_current_allocated_memory = 0;
static ulint mem_max_allocated_memory = 0;
static ulint mem_last_print_info = 0;
/* Size of the hash table for memory management tracking */
#define MEM_HASH_SIZE 997
......
......@@ -25,7 +25,6 @@
#ifdef __cplusplus
#include <stdlib.h>
#include <unistd.h>
/* Use prototypes in function declarations. */
#define YY_USE_PROTOS
......@@ -934,31 +933,54 @@ case 3:
YY_RULE_SETUP
#line 116 "pars0lex.l"
{
/* Quoted character string literals are handled in an explicit
start state 'quoted'. This state is entered and the buffer for
the scanned string is emptied upon encountering a starting quote.
In the state 'quoted', only two actions are possible (defined below). */
BEGIN(quoted);
stringbuf_len = 0;
}
YY_BREAK
case 4:
YY_RULE_SETUP
#line 120 "pars0lex.l"
string_append(yytext, yyleng);
#line 125 "pars0lex.l"
{
/* Got a sequence of characters other than "'":
append to string buffer */
string_append(yytext, yyleng);
}
YY_BREAK
case 5:
YY_RULE_SETUP
#line 121 "pars0lex.l"
{ string_append(yytext, yyleng / 2);
#line 130 "pars0lex.l"
{
/* Got a sequence of "'" characters:
append half of them to string buffer,
as "''" represents a single "'".
We apply truncating division,
so that "'''" will result in "'". */
string_append(yytext, yyleng / 2);
/* If we got an odd number of quotes, then the
last quote we got is the terminating quote.
At the end of the string, we return to the
initial start state and report the scanned
string literal. */
if (yyleng % 2) {
BEGIN(INITIAL);
yylval = sym_tab_add_str_lit(
pars_sym_tab_global,
stringbuf, stringbuf_len);
(byte*) stringbuf, stringbuf_len);
return(PARS_STR_LIT);
}
}
YY_BREAK
case 6:
YY_RULE_SETUP
#line 131 "pars0lex.l"
#line 154 "pars0lex.l"
{
yylval = sym_tab_add_null_lit(pars_sym_tab_global);
......@@ -967,521 +989,521 @@ YY_RULE_SETUP
YY_BREAK
case 7:
YY_RULE_SETUP
#line 137 "pars0lex.l"
#line 160 "pars0lex.l"
{
/* Implicit cursor name */
yylval = sym_tab_add_str_lit(pars_sym_tab_global,
yytext, yyleng);
(byte*) yytext, yyleng);
return(PARS_SQL_TOKEN);
}
YY_BREAK
case 8:
YY_RULE_SETUP
#line 144 "pars0lex.l"
#line 167 "pars0lex.l"
{
return(PARS_AND_TOKEN);
}
YY_BREAK
case 9:
YY_RULE_SETUP
#line 148 "pars0lex.l"
#line 171 "pars0lex.l"
{
return(PARS_OR_TOKEN);
}
YY_BREAK
case 10:
YY_RULE_SETUP
#line 152 "pars0lex.l"
#line 175 "pars0lex.l"
{
return(PARS_NOT_TOKEN);
}
YY_BREAK
case 11:
YY_RULE_SETUP
#line 156 "pars0lex.l"
#line 179 "pars0lex.l"
{
return(PARS_PROCEDURE_TOKEN);
}
YY_BREAK
case 12:
YY_RULE_SETUP
#line 160 "pars0lex.l"
#line 183 "pars0lex.l"
{
return(PARS_IN_TOKEN);
}
YY_BREAK
case 13:
YY_RULE_SETUP
#line 164 "pars0lex.l"
#line 187 "pars0lex.l"
{
return(PARS_OUT_TOKEN);
}
YY_BREAK
case 14:
YY_RULE_SETUP
#line 168 "pars0lex.l"
#line 191 "pars0lex.l"
{
return(PARS_INT_TOKEN);
}
YY_BREAK
case 15:
YY_RULE_SETUP
#line 172 "pars0lex.l"
#line 195 "pars0lex.l"
{
return(PARS_INT_TOKEN);
}
YY_BREAK
case 16:
YY_RULE_SETUP
#line 176 "pars0lex.l"
#line 199 "pars0lex.l"
{
return(PARS_FLOAT_TOKEN);
}
YY_BREAK
case 17:
YY_RULE_SETUP
#line 180 "pars0lex.l"
#line 203 "pars0lex.l"
{
return(PARS_CHAR_TOKEN);
}
YY_BREAK
case 18:
YY_RULE_SETUP
#line 184 "pars0lex.l"
#line 207 "pars0lex.l"
{
return(PARS_IS_TOKEN);
}
YY_BREAK
case 19:
YY_RULE_SETUP
#line 188 "pars0lex.l"
#line 211 "pars0lex.l"
{
return(PARS_BEGIN_TOKEN);
}
YY_BREAK
case 20:
YY_RULE_SETUP
#line 192 "pars0lex.l"
#line 215 "pars0lex.l"
{
return(PARS_END_TOKEN);
}
YY_BREAK
case 21:
YY_RULE_SETUP
#line 196 "pars0lex.l"
#line 219 "pars0lex.l"
{
return(PARS_IF_TOKEN);
}
YY_BREAK
case 22:
YY_RULE_SETUP
#line 200 "pars0lex.l"
#line 223 "pars0lex.l"
{
return(PARS_THEN_TOKEN);
}
YY_BREAK
case 23:
YY_RULE_SETUP
#line 204 "pars0lex.l"
#line 227 "pars0lex.l"
{
return(PARS_ELSE_TOKEN);
}
YY_BREAK
case 24:
YY_RULE_SETUP
#line 208 "pars0lex.l"
#line 231 "pars0lex.l"
{
return(PARS_ELSIF_TOKEN);
}
YY_BREAK
case 25:
YY_RULE_SETUP
#line 212 "pars0lex.l"
#line 235 "pars0lex.l"
{
return(PARS_LOOP_TOKEN);
}
YY_BREAK
case 26:
YY_RULE_SETUP
#line 216 "pars0lex.l"
#line 239 "pars0lex.l"
{
return(PARS_WHILE_TOKEN);
}
YY_BREAK
case 27:
YY_RULE_SETUP
#line 220 "pars0lex.l"
#line 243 "pars0lex.l"
{
return(PARS_RETURN_TOKEN);
}
YY_BREAK
case 28:
YY_RULE_SETUP
#line 224 "pars0lex.l"
#line 247 "pars0lex.l"
{
return(PARS_SELECT_TOKEN);
}
YY_BREAK
case 29:
YY_RULE_SETUP
#line 228 "pars0lex.l"
#line 251 "pars0lex.l"
{
return(PARS_SUM_TOKEN);
}
YY_BREAK
case 30:
YY_RULE_SETUP
#line 232 "pars0lex.l"
#line 255 "pars0lex.l"
{
return(PARS_COUNT_TOKEN);
}
YY_BREAK
case 31:
YY_RULE_SETUP
#line 236 "pars0lex.l"
#line 259 "pars0lex.l"
{
return(PARS_DISTINCT_TOKEN);
}
YY_BREAK
case 32:
YY_RULE_SETUP
#line 240 "pars0lex.l"
#line 263 "pars0lex.l"
{
return(PARS_FROM_TOKEN);
}
YY_BREAK
case 33:
YY_RULE_SETUP
#line 244 "pars0lex.l"
#line 267 "pars0lex.l"
{
return(PARS_WHERE_TOKEN);
}
YY_BREAK
case 34:
YY_RULE_SETUP
#line 248 "pars0lex.l"
#line 271 "pars0lex.l"
{
return(PARS_FOR_TOKEN);
}
YY_BREAK
case 35:
YY_RULE_SETUP
#line 252 "pars0lex.l"
#line 275 "pars0lex.l"
{
return(PARS_CONSISTENT_TOKEN);
}
YY_BREAK
case 36:
YY_RULE_SETUP
#line 256 "pars0lex.l"
#line 279 "pars0lex.l"
{
return(PARS_READ_TOKEN);
}
YY_BREAK
case 37:
YY_RULE_SETUP
#line 260 "pars0lex.l"
#line 283 "pars0lex.l"
{
return(PARS_ORDER_TOKEN);
}
YY_BREAK
case 38:
YY_RULE_SETUP
#line 264 "pars0lex.l"
#line 287 "pars0lex.l"
{
return(PARS_BY_TOKEN);
}
YY_BREAK
case 39:
YY_RULE_SETUP
#line 268 "pars0lex.l"
#line 291 "pars0lex.l"
{
return(PARS_ASC_TOKEN);
}
YY_BREAK
case 40:
YY_RULE_SETUP
#line 272 "pars0lex.l"
#line 295 "pars0lex.l"
{
return(PARS_DESC_TOKEN);
}
YY_BREAK
case 41:
YY_RULE_SETUP
#line 276 "pars0lex.l"
#line 299 "pars0lex.l"
{
return(PARS_INSERT_TOKEN);
}
YY_BREAK
case 42:
YY_RULE_SETUP
#line 280 "pars0lex.l"
#line 303 "pars0lex.l"
{
return(PARS_INTO_TOKEN);
}
YY_BREAK
case 43:
YY_RULE_SETUP
#line 284 "pars0lex.l"
#line 307 "pars0lex.l"
{
return(PARS_VALUES_TOKEN);
}
YY_BREAK
case 44:
YY_RULE_SETUP
#line 288 "pars0lex.l"
#line 311 "pars0lex.l"
{
return(PARS_UPDATE_TOKEN);
}
YY_BREAK
case 45:
YY_RULE_SETUP
#line 292 "pars0lex.l"
#line 315 "pars0lex.l"
{
return(PARS_SET_TOKEN);
}
YY_BREAK
case 46:
YY_RULE_SETUP
#line 296 "pars0lex.l"
#line 319 "pars0lex.l"
{
return(PARS_DELETE_TOKEN);
}
YY_BREAK
case 47:
YY_RULE_SETUP
#line 300 "pars0lex.l"
#line 323 "pars0lex.l"
{
return(PARS_CURRENT_TOKEN);
}
YY_BREAK
case 48:
YY_RULE_SETUP
#line 304 "pars0lex.l"
#line 327 "pars0lex.l"
{
return(PARS_OF_TOKEN);
}
YY_BREAK
case 49:
YY_RULE_SETUP
#line 308 "pars0lex.l"
#line 331 "pars0lex.l"
{
return(PARS_CREATE_TOKEN);
}
YY_BREAK
case 50:
YY_RULE_SETUP
#line 312 "pars0lex.l"
#line 335 "pars0lex.l"
{
return(PARS_TABLE_TOKEN);
}
YY_BREAK
case 51:
YY_RULE_SETUP
#line 316 "pars0lex.l"
#line 339 "pars0lex.l"
{
return(PARS_INDEX_TOKEN);
}
YY_BREAK
case 52:
YY_RULE_SETUP
#line 320 "pars0lex.l"
#line 343 "pars0lex.l"
{
return(PARS_UNIQUE_TOKEN);
}
YY_BREAK
case 53:
YY_RULE_SETUP
#line 324 "pars0lex.l"
#line 347 "pars0lex.l"
{
return(PARS_CLUSTERED_TOKEN);
}
YY_BREAK
case 54:
YY_RULE_SETUP
#line 328 "pars0lex.l"
#line 351 "pars0lex.l"
{
return(PARS_DOES_NOT_FIT_IN_MEM_TOKEN);
}
YY_BREAK
case 55:
YY_RULE_SETUP
#line 332 "pars0lex.l"
#line 355 "pars0lex.l"
{
return(PARS_ON_TOKEN);
}
YY_BREAK
case 56:
YY_RULE_SETUP
#line 336 "pars0lex.l"
#line 359 "pars0lex.l"
{
return(PARS_DECLARE_TOKEN);
}
YY_BREAK
case 57:
YY_RULE_SETUP
#line 340 "pars0lex.l"
#line 363 "pars0lex.l"
{
return(PARS_CURSOR_TOKEN);
}
YY_BREAK
case 58:
YY_RULE_SETUP
#line 344 "pars0lex.l"
#line 367 "pars0lex.l"
{
return(PARS_OPEN_TOKEN);
}
YY_BREAK
case 59:
YY_RULE_SETUP
#line 348 "pars0lex.l"
#line 371 "pars0lex.l"
{
return(PARS_FETCH_TOKEN);
}
YY_BREAK
case 60:
YY_RULE_SETUP
#line 352 "pars0lex.l"
#line 375 "pars0lex.l"
{
return(PARS_CLOSE_TOKEN);
}
YY_BREAK
case 61:
YY_RULE_SETUP
#line 356 "pars0lex.l"
#line 379 "pars0lex.l"
{
return(PARS_NOTFOUND_TOKEN);
}
YY_BREAK
case 62:
YY_RULE_SETUP
#line 360 "pars0lex.l"
#line 383 "pars0lex.l"
{
return(PARS_TO_CHAR_TOKEN);
}
YY_BREAK
case 63:
YY_RULE_SETUP
#line 364 "pars0lex.l"
#line 387 "pars0lex.l"
{
return(PARS_TO_NUMBER_TOKEN);
}
YY_BREAK
case 64:
YY_RULE_SETUP
#line 368 "pars0lex.l"
#line 391 "pars0lex.l"
{
return(PARS_TO_BINARY_TOKEN);
}
YY_BREAK
case 65:
YY_RULE_SETUP
#line 372 "pars0lex.l"
#line 395 "pars0lex.l"
{
return(PARS_BINARY_TO_NUMBER_TOKEN);
}
YY_BREAK
case 66:
YY_RULE_SETUP
#line 376 "pars0lex.l"
#line 399 "pars0lex.l"
{
return(PARS_SUBSTR_TOKEN);
}
YY_BREAK
case 67:
YY_RULE_SETUP
#line 380 "pars0lex.l"
#line 403 "pars0lex.l"
{
return(PARS_REPLSTR_TOKEN);
}
YY_BREAK
case 68:
YY_RULE_SETUP
#line 384 "pars0lex.l"
#line 407 "pars0lex.l"
{
return(PARS_CONCAT_TOKEN);
}
YY_BREAK
case 69:
YY_RULE_SETUP
#line 388 "pars0lex.l"
#line 411 "pars0lex.l"
{
return(PARS_INSTR_TOKEN);
}
YY_BREAK
case 70:
YY_RULE_SETUP
#line 392 "pars0lex.l"
#line 415 "pars0lex.l"
{
return(PARS_LENGTH_TOKEN);
}
YY_BREAK
case 71:
YY_RULE_SETUP
#line 396 "pars0lex.l"
#line 419 "pars0lex.l"
{
return(PARS_SYSDATE_TOKEN);
}
YY_BREAK
case 72:
YY_RULE_SETUP
#line 400 "pars0lex.l"
#line 423 "pars0lex.l"
{
return(PARS_PRINTF_TOKEN);
}
YY_BREAK
case 73:
YY_RULE_SETUP
#line 404 "pars0lex.l"
#line 427 "pars0lex.l"
{
return(PARS_ASSERT_TOKEN);
}
YY_BREAK
case 74:
YY_RULE_SETUP
#line 408 "pars0lex.l"
#line 431 "pars0lex.l"
{
return(PARS_RND_TOKEN);
}
YY_BREAK
case 75:
YY_RULE_SETUP
#line 412 "pars0lex.l"
#line 435 "pars0lex.l"
{
return(PARS_RND_STR_TOKEN);
}
YY_BREAK
case 76:
YY_RULE_SETUP
#line 416 "pars0lex.l"
#line 439 "pars0lex.l"
{
return(PARS_ROW_PRINTF_TOKEN);
}
YY_BREAK
case 77:
YY_RULE_SETUP
#line 420 "pars0lex.l"
#line 443 "pars0lex.l"
{
return(PARS_COMMIT_TOKEN);
}
YY_BREAK
case 78:
YY_RULE_SETUP
#line 424 "pars0lex.l"
#line 447 "pars0lex.l"
{
return(PARS_ROLLBACK_TOKEN);
}
YY_BREAK
case 79:
YY_RULE_SETUP
#line 428 "pars0lex.l"
#line 451 "pars0lex.l"
{
return(PARS_WORK_TOKEN);
}
YY_BREAK
case 80:
YY_RULE_SETUP
#line 432 "pars0lex.l"
#line 455 "pars0lex.l"
{
yylval = sym_tab_add_id(pars_sym_tab_global,
(byte*)yytext,
......@@ -1491,42 +1513,42 @@ YY_RULE_SETUP
YY_BREAK
case 81:
YY_RULE_SETUP
#line 439 "pars0lex.l"
#line 462 "pars0lex.l"
{
return(PARS_DDOT_TOKEN);
}
YY_BREAK
case 82:
YY_RULE_SETUP
#line 443 "pars0lex.l"
#line 466 "pars0lex.l"
{
return(PARS_ASSIGN_TOKEN);
}
YY_BREAK
case 83:
YY_RULE_SETUP
#line 447 "pars0lex.l"
#line 470 "pars0lex.l"
{
return(PARS_LE_TOKEN);
}
YY_BREAK
case 84:
YY_RULE_SETUP
#line 451 "pars0lex.l"
#line 474 "pars0lex.l"
{
return(PARS_GE_TOKEN);
}
YY_BREAK
case 85:
YY_RULE_SETUP
#line 455 "pars0lex.l"
#line 478 "pars0lex.l"
{
return(PARS_NE_TOKEN);
}
YY_BREAK
case 86:
YY_RULE_SETUP
#line 459 "pars0lex.l"
#line 482 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1534,7 +1556,7 @@ YY_RULE_SETUP
YY_BREAK
case 87:
YY_RULE_SETUP
#line 464 "pars0lex.l"
#line 487 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1542,7 +1564,7 @@ YY_RULE_SETUP
YY_BREAK
case 88:
YY_RULE_SETUP
#line 469 "pars0lex.l"
#line 492 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1550,7 +1572,7 @@ YY_RULE_SETUP
YY_BREAK
case 89:
YY_RULE_SETUP
#line 474 "pars0lex.l"
#line 497 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1558,7 +1580,7 @@ YY_RULE_SETUP
YY_BREAK
case 90:
YY_RULE_SETUP
#line 479 "pars0lex.l"
#line 502 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1566,7 +1588,7 @@ YY_RULE_SETUP
YY_BREAK
case 91:
YY_RULE_SETUP
#line 484 "pars0lex.l"
#line 507 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1574,7 +1596,7 @@ YY_RULE_SETUP
YY_BREAK
case 92:
YY_RULE_SETUP
#line 489 "pars0lex.l"
#line 512 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1582,7 +1604,7 @@ YY_RULE_SETUP
YY_BREAK
case 93:
YY_RULE_SETUP
#line 494 "pars0lex.l"
#line 517 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1590,7 +1612,7 @@ YY_RULE_SETUP
YY_BREAK
case 94:
YY_RULE_SETUP
#line 499 "pars0lex.l"
#line 522 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1598,7 +1620,7 @@ YY_RULE_SETUP
YY_BREAK
case 95:
YY_RULE_SETUP
#line 504 "pars0lex.l"
#line 527 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1606,7 +1628,7 @@ YY_RULE_SETUP
YY_BREAK
case 96:
YY_RULE_SETUP
#line 509 "pars0lex.l"
#line 532 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1614,7 +1636,7 @@ YY_RULE_SETUP
YY_BREAK
case 97:
YY_RULE_SETUP
#line 514 "pars0lex.l"
#line 537 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1622,7 +1644,7 @@ YY_RULE_SETUP
YY_BREAK
case 98:
YY_RULE_SETUP
#line 519 "pars0lex.l"
#line 542 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1630,7 +1652,7 @@ YY_RULE_SETUP
YY_BREAK
case 99:
YY_RULE_SETUP
#line 524 "pars0lex.l"
#line 547 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1638,7 +1660,7 @@ YY_RULE_SETUP
YY_BREAK
case 100:
YY_RULE_SETUP
#line 529 "pars0lex.l"
#line 552 "pars0lex.l"
{
return((int)(*yytext));
......@@ -1646,32 +1668,32 @@ YY_RULE_SETUP
YY_BREAK
case 101:
YY_RULE_SETUP
#line 534 "pars0lex.l"
#line 557 "pars0lex.l"
BEGIN(comment); /* eat up comment */
YY_BREAK
case 102:
YY_RULE_SETUP
#line 536 "pars0lex.l"
#line 559 "pars0lex.l"
YY_BREAK
case 103:
YY_RULE_SETUP
#line 537 "pars0lex.l"
#line 560 "pars0lex.l"
YY_BREAK
case 104:
YY_RULE_SETUP
#line 538 "pars0lex.l"
#line 561 "pars0lex.l"
BEGIN(INITIAL);
YY_BREAK
case 105:
YY_RULE_SETUP
#line 540 "pars0lex.l"
#line 563 "pars0lex.l"
/* eat up whitespace */
YY_BREAK
case 106:
YY_RULE_SETUP
#line 543 "pars0lex.l"
#line 566 "pars0lex.l"
{
fprintf(stderr,"Unrecognized character: %02x\n",
*yytext);
......@@ -1683,10 +1705,10 @@ YY_RULE_SETUP
YY_BREAK
case 107:
YY_RULE_SETUP
#line 552 "pars0lex.l"
#line 575 "pars0lex.l"
YY_FATAL_ERROR( "flex scanner jammed" );
YY_BREAK
#line 1687 "lex.yy.c"
#line 1710 "lex.yy.c"
case YY_STATE_EOF(INITIAL):
case YY_STATE_EOF(comment):
case YY_STATE_EOF(quoted):
......@@ -2574,5 +2596,5 @@ int main()
return 0;
}
#endif
#line 552 "pars0lex.l"
#line 575 "pars0lex.l"
......@@ -114,11 +114,34 @@ ID [a-z_A-Z][a-z_A-Z0-9]*
}
"'" {
/* Quoted character string literals are handled in an explicit
start state 'quoted'. This state is entered and the buffer for
the scanned string is emptied upon encountering a starting quote.
In the state 'quoted', only two actions are possible (defined below). */
BEGIN(quoted);
stringbuf_len = 0;
}
<quoted>[^\']+ string_append(yytext, yyleng);
<quoted>"'"+ { string_append(yytext, yyleng / 2);
<quoted>[^\']+ {
/* Got a sequence of characters other than "'":
append to string buffer */
string_append(yytext, yyleng);
}
<quoted>"'"+ {
/* Got a sequence of "'" characters:
append half of them to string buffer,
as "''" represents a single "'".
We apply truncating division,
so that "'''" will result in "'". */
string_append(yytext, yyleng / 2);
/* If we got an odd number of quotes, then the
last quote we got is the terminating quote.
At the end of the string, we return to the
initial start state and report the scanned
string literal. */
if (yyleng % 2) {
BEGIN(INITIAL);
yylval = sym_tab_add_str_lit(
......
.TH mysqlaccess 1 "19 December 2000" "MySQL @MYSQL_BASE_VERSION@" "MySQL database"
.SH NAME
.BR mysqlaccess
\- Create new users to mysql.
.BR mysqlaccess \- Create new users to mysql.
.SH USAGE
mysqlaccess [host [user [db]]] OPTIONS
.SH SYNOPSIS
......
.TH mysqldump 1 "19 December 2000" "MySQL @MYSQL_BASE_VERSION@" "MySQL database"
.SH NAME
mysqldump \- text-based client for dumping or backing up mysql databases , tables and or data.
mysqldump \- text\-based client for dumping or backing up mysql databases, tables and or data.
.SH USAGE
.BR "mysqldump [\fP\fIOPTIONS\fP] database [\fP\fItables\fP]"
......
......@@ -98,7 +98,7 @@ commit;
show status like "Qcache_queries_in_cache";
Variable_name Value
Qcache_queries_in_cache 1
drop table if exists t1;
drop table t3,t2,t1;
CREATE TABLE t1 (id int(11) NOT NULL auto_increment, PRIMARY KEY (id)) TYPE=InnoDB;
select count(*) from t1;
count(*)
......@@ -108,3 +108,22 @@ select count(*) from t1;
count(*)
1
drop table t1;
set GLOBAL query_cache_size=1355776;
CREATE TABLE t1 ( id int(10) NOT NULL auto_increment, a varchar(25) default NULL, PRIMARY KEY (id), UNIQUE KEY a (a)) TYPE=innodb;
CREATE TABLE t2 ( id int(10) NOT NULL auto_increment, b varchar(25) default NULL, PRIMARY KEY (id), UNIQUE KEY b (b)) TYPE=innodb;
CREATE TABLE t3 ( id int(10) NOT NULL auto_increment, t1_id int(10) NOT NULL default '0', t2_id int(10) NOT NULL default '0', state int(11) default NULL, PRIMARY KEY (id), UNIQUE KEY t1_id (t1_id,t2_id), KEY t2_id (t2_id,t1_id), CONSTRAINT `t3_ibfk_1` FOREIGN KEY (`t1_id`) REFERENCES `t1` (`id`), CONSTRAINT `t3_ibfk_2` FOREIGN KEY (`t2_id`) REFERENCES `t2` (`id`)) TYPE=innodb;
INSERT INTO t1 VALUES (1,'me');
INSERT INTO t2 VALUES (1,'you');
INSERT INTO t3 VALUES (2,1,1,2);
delete from t3 where t1_id = 1 and t2_id = 1;
select t1.* from t1, t2, t3 where t3.state & 1 = 0 and t3.t1_id = t1.id and t3.t2_id = t2.id and t1.id = 1 order by t1.a asc;
id a
begin;
insert into t3 VALUES ( NULL, 1, 1, 2 );
insert into t3 VALUES ( NULL, 1, 1, 2 );
Duplicate entry '1-1' for key 2
commit;
select t1.* from t1, t2, t3 where t3.state & 1 = 0 and t3.t1_id = t1.id and t3.t2_id = t2.id and t1.id = 1 order by t1.a asc;
id a
1 me
drop table t3,t2,t1;
......@@ -255,7 +255,7 @@ t1 ref y y 5 const 1 Using where
t2 range x x 5 NULL 2 Using where
explain select count(*) from t1 where x in (1);
table type possible_keys key key_len ref rows Extra
t1 range x x 5 NULL 1 Using where; Using index
t1 ref x x 5 const 1 Using where; Using index
explain select count(*) from t1 where x in (1,2);
table type possible_keys key key_len ref rows Extra
t1 range x x 5 NULL 2 Using where; Using index
......
......@@ -48,10 +48,31 @@ show status like "Qcache_queries_in_cache";
show status like "Qcache_hits";
commit;
show status like "Qcache_queries_in_cache";
drop table t3,t2,t1;
drop table if exists t1;
CREATE TABLE t1 (id int(11) NOT NULL auto_increment, PRIMARY KEY (id)) TYPE=InnoDB;
select count(*) from t1;
insert into t1 (id) values (0);
select count(*) from t1;
drop table t1;
#
# one statement roll back inside transation
#
set GLOBAL query_cache_size=1355776;
CREATE TABLE t1 ( id int(10) NOT NULL auto_increment, a varchar(25) default NULL, PRIMARY KEY (id), UNIQUE KEY a (a)) TYPE=innodb;
CREATE TABLE t2 ( id int(10) NOT NULL auto_increment, b varchar(25) default NULL, PRIMARY KEY (id), UNIQUE KEY b (b)) TYPE=innodb;
CREATE TABLE t3 ( id int(10) NOT NULL auto_increment, t1_id int(10) NOT NULL default '0', t2_id int(10) NOT NULL default '0', state int(11) default NULL, PRIMARY KEY (id), UNIQUE KEY t1_id (t1_id,t2_id), KEY t2_id (t2_id,t1_id), CONSTRAINT `t3_ibfk_1` FOREIGN KEY (`t1_id`) REFERENCES `t1` (`id`), CONSTRAINT `t3_ibfk_2` FOREIGN KEY (`t2_id`) REFERENCES `t2` (`id`)) TYPE=innodb;
INSERT INTO t1 VALUES (1,'me');
INSERT INTO t2 VALUES (1,'you');
INSERT INTO t3 VALUES (2,1,1,2);
delete from t3 where t1_id = 1 and t2_id = 1;
select t1.* from t1, t2, t3 where t3.state & 1 = 0 and t3.t1_id = t1.id and t3.t2_id = t2.id and t1.id = 1 order by t1.a asc;
begin;
insert into t3 VALUES ( NULL, 1, 1, 2 );
-- error 1062
insert into t3 VALUES ( NULL, 1, 1, 2 );
commit;
select t1.* from t1, t2, t3 where t3.state & 1 = 0 and t3.t1_id = t1.id and t3.t2_id = t2.id and t1.id = 1 order by t1.a asc;
drop table t3,t2,t1;
......@@ -182,7 +182,8 @@ uint calc_hashnr_caseup(const byte *key, uint len)
#endif
#ifndef __SUNPRO_C /* SUNPRO can't handle this */
/* for compilers which can not handle inline */
#if !defined(__SUNPRO_C) && !defined(__USLC__) && !defined(__sgi)
inline
#endif
unsigned int rec_hashnr(HASH *hash,const byte *record)
......
......@@ -446,13 +446,12 @@ int ha_rollback_trans(THD *thd, THD_TRANS *trans)
reinit_io_cache(&thd->transaction.trans_log,
WRITE_CACHE, (my_off_t) 0, 0, 1);
thd->transaction.trans_log.end_of_file= max_binlog_cache_size;
if (operation_done)
thd->transaction.cleanup();
}
thd->variables.tx_isolation=thd->session_tx_isolation;
if (operation_done)
{
statistic_increment(ha_rollback_count,&LOCK_status);
thd->transaction.cleanup();
}
}
#endif /* USING_TRANSACTIONS */
DBUG_RETURN(error);
......
......@@ -1590,12 +1590,14 @@ add_key_fields(JOIN_TAB *stat,KEY_FIELD **key_fields,uint *and_level,
// BETWEEN or IN
if (cond_func->key_item()->type() == Item::FIELD_ITEM)
add_key_field(key_fields,*and_level,
((Item_field*) (cond_func->key_item()))->field, 0,
((Item_field*) (cond_func->key_item()))->field,
#ifndef TO_BE_REMOVED_IN_4_1
/* special treatment for IN. Not necessary in 4.1 */
cond_func->argument_count() == 1,
cond_func->arguments() + (cond_func->functype() != Item_func::IN_FUNC),
cond_func->argument_count() - (cond_func->functype() != Item_func::IN_FUNC),
#else
cond_func->argument_count() == 2,
cond_func->arguments()+1, cond_func->argument_count()-1,
#endif
usable_tables);
......
......@@ -91,7 +91,12 @@ main(int argc __attribute__((unused)), char** argv)
struct sockaddr_in sa_cli;
int listen_sd;
int err;
#if defined(__sgi) && _NO_XOPEN4 && _NO_XOPEN5
socklen_t client_len;
#else
size_t client_len;
#endif
int reuseaddr = 1; /* better testing, uh? */
MY_INIT(argv[0]);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment