- Fixes added in CegoAction and CegoSelect to treat cego parser
with tablemanager pointer = 0.
This is needed for cegobridge to parse procedures and generated
other db specific code
- Fix added in CegoObjectManager::alterObject, after freeing old page entry,
page should be unfixed with marked as dirty. In the main loop,
unfix has been changed to dirty = false
- Added function dateformat which can be feeded with two expressions
to return a datetime value from a given format and value string (
analog to date(x,y) which expects constant input values )
- Added more date functions to the datefunc package. Now there are
the following date functions available
int second(datetime d)
int minute(datetime d)
int hour(datetime d)
int day(datetime d)
int month(datetime d)
int year(datetime d)
datetime plusHour(datetime d, int hr)
datetime plusDay(datetime d, int dy)
datetime plusMonth(datetime d, int mt)
datetime plusYear(datetime d, int yr)
datetime minusHour(datetime d, int hr)
datetime minusDay(datetime d, int dy)
datetime minusMonth(datetime d, int mt)
datetime minusYear(datetime d, int yr)
- In CegoMain, added DEFAULTDATETIMEFORMAT1 to dateFormat list to
also parse this format for native cast from string to datetime
- Added Query Cache menu to CegoAdmScreen to manage query cache parameters
- Fix for some geometries in Screen::showFormBox
- Further improvements for Screen::showFormBox and showInfoBox
( selected rows and newline treatment )
- integrate Screen2 interface
databases/cego: update 2.48.2 -> 2.48.10
- Introduced Backup branch id for online backup verification. Due
to datafile page id constraints, just online backups can be recovered
inside one backup branch. Creation of a tableset and adding a
datafile to a tableset violates this contraint, so a new backup
branch is created.
- Added backup branch option for backup manager. So just valid
backup files can be treated by the manager. New sample backup manager
available in samples/chkdb
- Further improvements for cgadm screen mode. confirm dialogs added
and backup log history indicated for online backups in case of errors
- More improvements for online backup / restore. Branch id is now
added to archive log filename, so in case of new created or data
file modified tablesets, the corresponding logfiles can be filtered.
Archive logfile information for ready-for-archive and archived
logfiles has been added to archive log information
- Fixes added in CegoClient for dump mode. Some attribute and
attribute values have not treated correctly ( default values,
dimension for fixed attributes, clob values )
- Fixes added for CegoClient dump and batch mode, has been tested
with more complex acc application
- Introduced grace mode for graceful object creation. Using this
mode, views and procedures can refer to objects, which still not
exist. This mode might be useful for initial loads ( e.g. from a
dump file ) to avoid dependency errors
- Fix for CegoSelect::nextTuple, _cacheSchema was not setup for
union selects in case of empty result of the first union entry.
This leads to an exception with cache claim leak. The claim leak
has also been fixed by adding a cleanUp call to CegoSelect destructor
- Fix added in CegoAction::execViewCreate(), in case of grace mode,
empty schema must be stored with created view to force recompile
via CegoDistManager::reloadView
- Added defTabSetId parameter to several toChain methods to write
views, procedures and triggers without explizit tableset references.
This might be useful, if a tableset dump just references objects
for a single tableset and it should be loaded to a tableset with a
different name. ( e.g. select a from t1@TS1 is written as select a
from t1, if active tableset is TS1 )
- Fixes added for alias object usage ( e.g. in combination with views )
- Added lfcbase Screen2 support to CegoAdmScreen
databases/cegobridge: update 1.6.2 -> 1.6.3
- Build fixes added
Author: Björn Lemke <lemke@lemke-it.com>
lfcbase:
- Improvements done for ncurses based Monitor class.
Added confirm box and improved form box
- many changed have been made to Screen class ( was Monitor ).
The changes require a recompile for consuming packages ( here : cego )
- Improvements done for Screen::showFormBox ( scrolling content in input mask )
cego:
- Fix added in CegoFactor and CegoFieldValue.
To get the factor id for query cache management, a dedicated
method was introduced in CegoFieldValue ( getId ) This method
uses for DATETIME values valAsChain(true) ( to evaluate current
timestamp in case of sysdate values ) and for all others toChain()
( to distinguish x = null and x = 'null' )
- Fix in CegoFunction::getId, the GETCOUNT function still was not
treated
- Some cleanup of source file was done (removed tilde files)
- Added join optimization fix in CegoSelect::buildJoinTables.
Inadequate predicates ( e.g. not equal compare ) are not priorized
for table join order any more. For this the method
CegoPredDesc::prioOnJoin has been introduced
- Grammar fix in Cego.def to ensure _returnOnFirst variable
is set to false for non return update statements
( added production NoUpdateReturnOpt )
- Fix added in CegoTableManager::updateTuple,
for the given expression list, contained subqueries must be
retrieved and analyzed for external table references.
Since the external count was not evaluated correctly, this led
to a incomplete CegoSelect::getQueryId result.
With enabled query cache, the following update did not work correct:
UPDATE posrule pr SET artid =
( SELECT artid FROM article a WHERE a.artnr = pr.artnr );
- Fix added in CegoXMLSpace for user and role create routines.
In case of already defined entities, exception was thrown
before xml lock release
Fix added in CegoXMLSpace::getActiveTableSet to also retrieve
offline tablesets via includeOffline flag.
This is needed for CegoLogThreadPool to archive occupied logfiles
for tablesets in status offline ( for other reasons, runstate is
set now at the beginning of the tableset stop procedure )
- many changes made in CegoAdmScreen class ( was CegoAdmMon )
Several stabilization fixes also have been added. For database
xml file, a unique database id has been introduced to validate
ticket file from performed online backups ( db id must match )
- Small adaptions for cgadm screen mode window sizes
( corresponds to lfcbase-1.17.2 )
Fix added in CegoTableManager::alterDataTable, the altering table
object has to be retrieved completely via getObject to setup
data page id and last data page id. With the new introduced
object manager method alterObject, these values
are not copied during the alter procedure.
Fix in CegoObjectManager for alterTableObject.
If a new system page has to be allocated, this page was not linked.
This might lead to loss table objects for corresponding altering
queries. Furthermore, the ObjectManager has been cleaned up and
reduced in code size in such a way that generic create and alter
methods are now provided for all objects
( createObject and alterObject using encoding method inherited from
CegoDecodableObject ). This results in less code, since the create/alter
operations are just implemented once.
lfcbase:
- Added new method BigDecimal::scaleTo to change scaling for an instance.
This includes an appropriate rounding, if scaling is reduced
cego:
- In CegoFieldValue::castTo, cast from varchar to fixed is done
now with newly introduced BigDecimal::scaleTo method
This includes an appropriate rounding in case of precision reduction.
- Fix in CegoCaseCondition to allow null types in condition
see dbcheck/check106.sql for a sample
Commit b7f05445c0 has added WWW entries to port Makefiles based on
WWW: lines in pkg-descr files.
This commit removes the WWW: lines of moved-over URLs from these
pkg-descr files.
Approved by: portmgr (tcberner)
It has been common practice to have one or more URLs at the end of the
ports' pkg-descr files, one per line and prefixed with "WWW:". These
URLs should point at a project website or other relevant resources.
Access to these URLs required processing of the pkg-descr files, and
they have often become stale over time. If more than one such URL was
present in a pkg-descr file, only the first one was tarnsfered into
the port INDEX, but for many ports only the last line did contain the
port specific URL to further information.
There have been several proposals to make a project URL available as
a macro in the ports' Makefiles, over time.
This commit implements such a proposal and moves one of the WWW: entries
of each pkg-descr file into the respective port's Makefile. A heuristic
attempts to identify the most relevant URL in case there is more than
one WWW: entry in some pkg-descr file. URLs that are not moved into the
Makefile are prefixed with "See also:" instead of "WWW:" in the pkg-descr
files in order to preserve them.
There are 1256 ports that had no WWW: entries in pkg-descr files. These
ports will not be touched in this commit.
The portlint port has been adjusted to expect a WWW entry in each port
Makefile, and to flag any remaining "WWW:" lines in pkg-descr files as
deprecated.
Approved by: portmgr (tcberner)
lfcbase:
- Fix in BigDecimal::operator ==, if _val, _scale and _isPositive
do not match, we still have to call further checks
cego:
- Fix in CegoProcBlock::toChain, dimension has still not been
treated for fixed procedure variables
lfcbase:
- Fix added for BigDecimal to avoid rounding for zero trailing values,
e.g. 119.00 / 100.00 was 1.1, should be 1.19
cego:
- Added small patch for verification check to meet new BigDecimal patch
in lfcbase
- Replaced Chain("") with Chain() for several. This was motivated by
an observed bug for check constraint xml import, where a comparision
between Chain() and Chain("") returned false
- Fix in CegoFunction::evalFieldValue for case REPLACE. If replace
string is null, then empty string is replaced instead of "null"
string.
Author: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoFunction::clone, also if _exprList is empty, the function
must treated and cloned completely including table manager reference.
Otherwise, for user defined function with no arguments which are used
in views will throw an exception
- Change in CegoJDBCInfo for table list retrievel. This JDBC meta
information command is not split to tab and vie for dedicated
retrievel of table and view objects
- More work has been done regarding clean shutdown of the
database / tableset stop while busy db sessions still exist.
For this, CegoDistCursr desctructor method has been stabilized,
so no exception is thrown. Otherwise, double memory freeing of
cursor resources might occur which results in a crash
- In CegoTableManager::stopTableset, relocated the setTableSetRunState
method above and added a sleep of DBM_LOCKDELAY. This gives hanging
db threads a chance, to terminate save
- Changed CegoDatabaseManager::useObject method to take pointer to
CegoTableManager as an argument instead of threadId. The tablemanager
is used to support thread termination in case of lock delays
- Fix in CegoTableManager::createBTree to catch exclusicve useObject
timeout before committing btree. In this case, the already created
btree object should be deleted. Fix added in CegoClient to active
abort interrupt also during tuple fetch
- Added function getcount to retrieve current counter value for
a specific counter
- Fix in CegoLogManager::stopLog, the log handle must NOT be deleted,
otherwise, a subsequent startLog and logAction method will result
in a dump. This effect has been observed while importing a single
table ( via xml import ) which failed, if the table already exists.
In this case, no checkpoint is written and log handle i no more
valid. A subsequent modifying db request then results in a core
dump.
- Fix in CegoAttrComp for equal, lt and gt operators.
The compMode ATTR was not treated correctly in a way, that the
attrDesc part was not evaluated. This might lead to missing attr
cond entries for sophisticated queries.
A verify check ( verify008 ) has been added to check this case
Author: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoDistManager::renameLocalObject, before renaming the object,
it has to be checked, if the new object name is already in use
- Fix in CegoXPorter::readIndexObject, index schema expects tablename
instead of indexname.
As a result binary and plain imports with index objects have failed.
- Fix in CegoDistManager::startTableSet, the current lsn has
to be setup before performing the cleanup option. cleanup uses
bufferFix which might result in a checkpoint. For this the current
lsn has to be setup
- In CegoQuery, added execSchema which is copied for each execution
call from _schema ( to have a virgin schema definition ). The
execSchema then is enriched in the appropriate way. For the insert
query, an addExprList variable was decleare to store the addition
needed expression to complete the insert value list ( instead of
adding it to _exprListArray ). After performing the query, the
allocated and inserted expression have to be deleted. This
implementation seems to be more useful for procedure based queries,
where the intiantiated CegoQuery object is executed multiple times
- In CegoDistManager, reload message for views, procedure and trigger
is avoided now ( since this occurs for each db thread during startup )
- In CegoAttrComp::setup(ListT** pJoinBuf, int offset) the _isSetup
flag was not set to true for the posSetup case. This might lead to
invalid error message. See https://www.lemke-it.com/litexec?request=pubnews
for details
- Fix in CegoGroupSpace::getValueForAgg, The return value
( of type CegoFieldValue ) must be converted to a local copy,
since the reference to the bufferpool might be invalidated
- Added format correction in CegoField::castTo for case VARCHAR to
FIXED. In case of string values without leading zeros ( e.g. .345),
leading zero is concated.
Author: Björn Lemke <lemke@lemke-it.com>
lfcbase:
- Added HashT::getRangePos and HashT::RemovePos for more efficient
hash array manipulation
cego:
- Fix in CegoQueryCacheEntry::cleanCache, reset of all relevant
variables. Furthermore, in CegoQueryCache methods, remove from
hasharray is checked for return code true
- Usage of new HashT methods ( getRangePos and RemovePos ) for
CegoQueryCache to remove entries from query cache array more efficient
- Fix in CegoFactor::evalFieldValue(), for case CegoFactor::QUERY,
the fetched value must be saved to a local copy, otherwise the
buffer could be overwritten by the subsequent nextTuple method call
to complete the query for query cache. As a result, wrong values
are returned in case of heavy buffer cache relocation.
Also update databases/cego: update 2.47.1 -> 2.47.2
cego:
- Fix in CegoQueryHelper::checkIndexForPredicate, wrong tabSetId
was used for getObjectListByTable, must be table specific
cegobridge:
- In method void MySQLAction::storeVarcharType, dimension value was
doubled, since MySQL calculates visible length for varchars and
cego length in bytes. So for multibyte character encodings, length
might exceed ...
- Rework of new hash array based query cache.
The new introduced methods HashT::FirstInRange and HashT::NextInRange
are used to find an appropriate hash array slot to replace entries
in an occupied array.
- Added cego command line mode setpwd to setup user password
for offline database
databases/cego: update 2.46.8 -> 2.47.0
lfcbase:
- Introduction of template class HashT
cego:
- Integration of HashT into QueryCache for faster access of cached
query entries
- Fix in CegoClient and CegoAction for comment parsing. Comment token
in string are treated now. For this, a new method
CegoQueryHelper::skipComment has been introduced
- Improvements done for query cache handling
( more support for nested queries, stabilization fixes )
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Added fix in CegoTableManager::createBTree
For btree value extraction, we also have to use
valueFromSchema(tableSchema, btreeSchema)
Otherwise, in case if inverted index attributes, the values are
not extraced correctly
Example:
create table t1 ( a int, b string(10));
insert into t1 values ( 1, 'alpha');
create btree b1 on t1(b,a); -- inverted, value for a would be
not setup correctly
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoObjectManager / CegoDatabaseManager added.
The checkTableSetRunState method has been moved to CegoDatabaseManager.
This ensures a single point of check definition.
In the previous version, tableset RECOVERY mode was not treated for
useObject, unuseObject and objectExists method calls which might lead to
crash recovery abort.
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoExpr::evalField, fixed dimension was still not handled
here for ADD and SUB cases
- Added error msg improvement in CegoProcBlock
- Added MAXDELETEPAGE parameter to show parameter admin command
- set default checkpoint value for new defined tablesets
- added function check in CegoFunction::evalFieldValue to ensure, this is
not a procedure without return value
- Fix in CegoTableManager for table altering. No type conversion
allowed with alter modify and no length shrinking
Submitted by: Björn Lemke <lemke@lemke-it.com>
devel/lfcxml: update 1.2.11 -> 1.3.0
databases/cego: update 2.45.35 -> 2.46.0
databases/cegobridge: update 1.5.4 -> 1.6.0
Please note: database updates from 2.45.x to version 2.46.x require
a xml tableset export and import, since table structure information
has been changed!
lfcbase:
- Added template class GraphT for dependency graph implementations
- Fix in BigInteger::mul and BigInteger::div, signing was not treated
( e.g. -12 * 12 => -144 ( was 144 ) )
lfcxml: get in sync with lfcbase
cego:
- Added fix in CegoViewObject::getId. Since it cannot be decided,
if the underlying query result has been changed, we have to
throw MOD_QUERY_ID
- Another query cache fix in CegoFunction:getId, rand and liob
functions have been added to throw MOD_QUERY_ID exception, since
for those function it can not be ad hoc decided, if the result is
modified
- Added type dimension information for CegoField class. This enables the
database to define length and dim information for fixed and decimal
data types. For this the syntax has been expanded. Instead of storing
RESERVED_BTREE_FLOATLEN for btrees on fixed or decimal columns,
the specified length information is used now.
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Added noinit option to cgadm tableset start command. With the
noinit option, the configured init file will not be processed after
tableset startup, otherwise this will be processed now ( if any
configured ) All output messages are written to main log file.
- Redirected output for CegoAction and CegoOutput added to print output
to logfile. This is useful in case of tableset init file, where
several SQL action could be performed
- Fix in CegoAction::execSetCounter, new value must be checked for
null value and if so, set counter to 0. The bug caused a seg fault ..
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Improved runstate handling in CegoDistManager
- Further fix in CegoPredDesc::getFieldList, case CegoPredDesc::NULLCOMP,
getFieldList method call added for pExpr1,
in CegoPredDesc::setFieldListArray method calls added for
_pC and _pNotPred. For this, setFieldListArray method has been
introduced in CegoCondDesc. This fixes should complete case
cond handling e.g. for aggregation queries.
check097 has been expanded for this
Submitted by: Björn Lemke <lemke@lemke-it.com>
- CegoObjectManager has been modified to check tableset run state
for several methods. For this, method
CegoObjectManager::checkTableSetRunState has been introduced.
- Added check in CegoDatabaseManager::objectExists to verify, if
corresponding tableset is online. So instead of "invalid object"
exception, a tableset offline exception is thrown, which might be
more useful
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Added case-cond support for aggregation queries
- Fix in CegoCaseCond::evalField, all expression are evaluated now
and are checked for type mismatch. Expressions must return the same
type, otherwise this might lead to value evaluation problems. This
fix also leads to fix in CegoFieldValue::getDim() and CegoFactor::evalField
since now getDim ist used for field length specification which is
also compared by CegoCaseCond::evalField
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoTableManager::insertDataTable, before call
getObjectListByTable, type must be checked ( must be CegoObject::TABLE
), otherwise ( for system objects ) not found exception arises.
This error may appear with the admin commands begin backup and end
back ( stat information is written to bustat system table )
- Added format fix to CegoAction::formatTableInfo, to treat very
long object names, max name length of all corresponding objects is
now calculated and output is printed in appropriate format
- Change in CegoAction::execProcCall and CegoAction::execFuncCall,
procedure/function arguments are evaluated and cast in this methods
now. This is needed, to support string to clob casting for procedure
arguments. As a consequence the CegoProcedure::execute method has
been changed to receive ListT instead of ListT.
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Typo fix in CegoTerm::evalField and CegoExpr::evalField
- More rework on CegoFieldValue for improved cast handling
( especially on fixed values )
Submitted by: Björn Lemke <lemke@lemke-it.com>
databases/cego: update 2.45.19 -> 2.45.23
lfcxml:
- Sync patch for MinGW package build
cego:
- Fix in CegoFieldValue::castTo, case VARCHAR_TYPE to FIXED_TYPE,
plain integers have not been accepted. check091 added for this
- Fix in CegoTerm::evalField and CegoExpr::evalField, type and
length calculation must correspond to cast logic in CegoFieldValue
( corresponding operators for mul, div, add and sub )
- Fix in CegoProcVar::setValue, a castTo has been added to cast to
appropriate target type
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoAction::insertArg to check for duplicate attribute
specification in insert statement, e.g.
INSERT INTO t1 ( a, a ) VALUES ...
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoClient for treating interactive input. Ctl-D input
lead to segmentation fault, if used in multiline statements.
- Patch in CegoAction::formatTableInfo. For size calculation of
btree objects, null value must be used instead of INVALID, otherwise,
wrong values are indicated during btree build
- Patch in CegoLogManager::switchLogFile, we first have to switch log
file and then change the status to OCCUPIED. Otherwise, the logfile
archiver ( CegoLogThreadPool::shiftRedoLogs ) might access occupied
logfile in parallel. Since this is a different thread and File::open
seems to be not thread safe, bad things might happen ...
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Fix in CegoBtreeManager for duplicate null handling. The corresponding
btree value has to be checked to be null. For this, the method
CegoBTreeValue::isNull has been introduced
- Fix in CegoFieldValue::negate, for FIXED_TYPE values, trailing
zeros have to be treated. Otherwise, select null - a from t1 where
a = 0.10 will result in 0.1
- Patch in 'update with return' clause. It seems to be useful, that
in combination with return, update should return after first matching
tuple providing the updated tuple in the return clause. So the
values for the tuple can be stored in procedure variables.
create table t1 ( a int, b string(30));
insert into t1 values ( 1, 'DONE');
insert into t1 values ( 2, 'DONE');
insert into t1 values ( 3, 'WAIT');
insert into t1 values ( 4, 'DONE');
insert into t1 values ( 5, 'WAIT');
update t1 set b = 'DONE' where b = 'WAIT' return :a = a;
Just the tuple with a = 3 is update and the value of a is returned
in the procedure variable :a
- For update with return statements, added the on first option.
Since the return variable can only store one return values, this
option specifies, if the update should return after first tuple or
if all matching tuples should be updated.
- In CegoMain, changed separator token for tsdef from ":;" to "=",
since for mingw configurations, absolute path handling was broken.
This resulted in adapations for mkdb and cgmkdb scripts.
- Fix in CegoObjectManager::getObjectListByTable, if object does
not exist, an exception is thrown now.
This resulted in a fix for CegoQueryHelper::checkIndexForPredicate,
where now the table alias has to be mapped to the correct physical
table name ( via coList as a new method argument )
- Fix in CegoAttrDesc::evalTableReferences to treat alias objects in joins
- Grammar expansion to support inner/left outer/right outer joins
without a condition. This might be useful for specific join order
to use appropriate table indices ( see dbcheck/check088.sql )
- Fix in CegoQueryHelper::checkIndexForPredicate, index evaluation
for alias objects was still not treated
- Fix in CegoDistCursor::getPlan and CegoAction::getJoinPlanString
to cover execution plan for alias objects
Submitted by: Björn Lemke <lemke@lemke-it.com>
- Fix in CegoBTreeManager, the allowDuplicateNull flag was not
treated any more. The duplicate handling has been moved from
CegoBtreeNode to CegoBtreeManager
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Fix in CegoSelect::setVIewCond, a view condition id has to be
calculated which is later used for getQueryId(). Otherwise, wrong
query cache entries are used in case of views with outer conditions
- Added method CegoBTreeNode::verifyLeafFull for full btree
verification in a sense, that each btree value is compared with the
corresponding table data entry via datapointer reference
- Improvement in CegoAdmAction for input handling. Used the same
methods for nextChar and friends as in CegoAction
- In CegoXMLSpace, changed access methods for mediator, primary and
secondary in a way, that the default value ist the current hostname.
In this case, no attribute values are set up. This reduces complexity
and avoids to change the entries manually, if the hostname changes
(do not edit the database xml without any deeper knowledge).
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
databases/cego: upgrade 2.45.5 -> 2.45.6
lfcbase:
- In configure.ac added check for darwin. This is required, since
for File::flush implemention, darwin rather requires a fcntl call
with option F_FULLFSYNC instead of fsync ( see OSX man page for
fsync )
cego:
- Added command line option --fsync to enable physical disk synchronisation
for logging and checkpointing. This options slows down database
significantly but ensures consistent data in case of an operating
system crash
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Fix in CegoQuery::execute for INSERTBYSELECT_QUERY case.
Since the selected tuples are clustered ( MAX_CLUSTERED_INSERT) we have
to create a local copy for the corresponding field values.
Otherwise, tuple information could be invalidated by relocated
buffer pool pages
- Fix in CegoFieldValue::fastComp, the castTo method calls for t1
and t2 have to be switched, since first ist has to be checked to
cast to native btree datatype ( used in CegoBTreeCursor::traceLog,
inRange and fullMatch )
example:
create table t1 ( a int, d datetime);
create btree b1 on t1(b);
select a from t1 where b = '12.07.2019'; -- should be casted
-- from string to datetime
- Code cleanup to avoid warnings with -Wswitch-enum
- In CegoClient, added input data check via File::hasData ( available
with lfcbase 1.14.0 ). Now, pipe input is autodetected, so the
command line argument "--pipe" has been removed
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Treated alias insert queries with empty schema, e.g.
insert into a1 values ( 1, 'XXX');
- Some patches added for alias management to allow alias defintions
as a subset of the original table, e.g.
create table t1 ( a int, b string(10), c string(30))
create alias a1 on t1 ( a as ax, b as bx);
This has required some fixes in CegoAliasObject::mapSchema and
CegoQueryHelper::mapFLA
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Added table alias feature to grammar definition. Table aliases
allow alternate table name and attribute name definitions for a
table ( create alias ... )
- Added classes CegoAliasObject and CegoAttrAlias for table alias handling
- Implemented base semantic actions for creating, listing and
describe alias objects ( still no usage )
- In CegoDistManager::registerObjects, added registration for fkey,
check, trigger and alias
- Added alias mapping for insert, update and delete queries (
CegoQueryHelper::mapAliasPredicate still to implement )
- Basic alias mapping for select works, here's a sample SQL script
drop if exists table t1;
list alias;
desc alias a1;
drop if exists alias a1;
create table t1 ( a int, b string(30));
create alias a1 on t1 ( a as ax, b as bx );
insert into a1 ( ax, bx ) values ( 1, 'XXX');
insert into a1 ( ax, bx ) values ( 2, 'YYY');
insert into a1 ( ax, bx ) values ( 3, 'ZZZ');
select ax from a1;
- More work on table alias handling. Now works with more sophisticated
select queries ( e.g. select ax from a1 aa where aa.ax = 1; ) For
this, some fixes had to be done in CegoContentObject, since the the
table alias attribute ( _tabAlias ) was not setup in constructor
and other methods
- Added alias and trigger xml export and import feature ( trigger
xport was still not implemented ) Fix in CegoDbThread::loadObjects
for trigger reload, instead of triggerList.Next(), triggerList.First()
was called, which result in an infinite loop in case of existing
trigger objects
- Added trigger and alias object handling to CegoXPorter binary
export and import
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Some improved exception handling in CegoTableManager::checkIntegrity
( more detail information about violated object )
- Code cleanup CegoTableManager, drop/create/alter operations inside
a transaction no more allowed
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Further recovery fix in CegoTableManager::insertDataTable In case
of an insert exception ( e.g. duplicate pkey ), a checkpoint is
forced now after deleteData to ensure file consistency
In CegoObjectManager::insertPageData a second checkpoint is forced
after data copy, if a previous checkpoint has occured. Otherwise
this insert might be ignored during recovery because of lower LSN
- Fix in CegoBTreeManager::checkDuplicate, the check for uflag (
returned CegoBTreeNode::valueExists ) to check for high value is
not enough if, entries have been deleted from node.
Instead, all subsequent pages have to be checked until a value higher is found
- Introduced CegoBufferPool::getCPCount() to return to current
checkpoint count. This method is used by CegoTableManager,
CegoObjectManager and CegoTransactionManager to decide, if a
consitency checkpoint is required
- Version passed checkCrashRecovery with the following parameters
NUMRUN=100
DOTRANSACTION=-dotransaction
ARANGE=10000
AINTERVAL=500
ACOUNT=30000
PRIMARY=primary
POOLSIZE=191
All relevant checkpoint situations have been observed over all run
Submitted by: Bjoern Lemke <lemke@lemke-it.com>
- Fix added in CegoTransactionManager::commitUpdate, btree must be
inserted with insertBTreeWithCommit
- Fix added in CegoTransactionManager::getCrashAffectedTables,
schema entry must be searched with Find(CegoField(Chain(),
SYS_RB_TABLE))
Submitted by: Bjoern Lemke <lemke@lemke-it.com>