changelog

有关MYSQL的开源码

	received.

2002-07-16  Murray Cumming  <[e-mail]>

	* Removed all "using namespace"s and added std:: prefix in lots
	of places. It is now more likely to build with the latest gcc.

2001-05-02 Murray Cumming <[e-mail]>
* Result_Use::operator=(Result_Use) now
  now returns something.

2001-04-27 Murray Cumming <[e-mail]>
* ^M characters removed from Makefile.am
  files. I think.

Version 1.7.5:

2001-03-30 Murray Cumming <[e-mail]>
* Header guards prefixed by MYSQLCPPAPI_.

2001-03-23 Murray Cumming <[e-mail]>
* FieldType::sql() now returns "NULL" for
  empty non-string values.

2001-02-10 Murray Cumming <[e-mail]>
* Fixed gcc 2.96 no newline at end of file
  warnings.
  
2000-09-03 Murray Cumming <[e-mail]>
* Changed COPYING from GPL to LGPL. Whoops.
  Released 1.7.2a

2000-09-01 Murray Cumming <[e-mail]>
* Released 1.7.2.

2000-08-29 Murray Cumming <[e-mail]>
* Connection methods now throw exception if the Connection
  isn't open.

2000-08-27 Murray Cumming <[e-mail]>
* Added get_TypeCategory() and get_TypeOptionals() to FieldType.

2000-08-25 Murray Cumming <[e-mail]>
* Renamed Result to Result_Store and ResUse to Result_Use.
  That should be clearer, though people will still need to
  read some documentation (the MySQL manual) to see what
  the difference is.
* Changed headers to .h instead of .hh. Yes, that's a
  useless change.
* Put Row and Query files in their own directories.

2000-08-10 Murray Cumming <[e-mail]>
* Changed Fields to derive from vector and store the field
  info instead of getting it from mysql on demand. Now a
  Fields object can be copied and stored without worrying
  about when the parent Result or Connection is destroyed.

2000-08-08 Murray Cumming <[e-mail]>
* Changed Connection::query() to Connection::create_Query().
* Added ex_Locked exception which is now thrown instead
  of ex_BadQuery when the connection is locked. ex_Locked
  is now thrown from inside lock() so there's no need for
  lock() to have a bool return. I don't see the point of
  this lock/unlock. 
* Put Result, ResNSel, and ResUse in query_results dir.
* Renamed ResNSel to Result_NoData. Made data protected
  and added get_ methods.
* ResUse: Removed duplicate methods with slightly different
  names.
* const_subscript_container and subscript_iterator:
  - Removed SizeType and DiffType template args. Only the
  default are ever used, and the iterator can get them from
  the element type.
  - Recoded operator overloads to reuse each other's code.
  - subscript_iterator::operator-> now returns a pointer.
  - default constructor now initializes member data.
* Removed FieldTypes and FieldNames. Fields contains the
  same data.
* Removed MutableRow and MutableRes. They weren't used
  and users might think that changing their data
  actually changed the database.

2000-07-31 Murray Cumming <[e-mail]>
* Connection: Added set_ methods and removed args from
  connect(). There are too many possible combinations of
  connect arguments, and this can not be solved
  unambigously with default args.
  Also removed constructors with args. It's OK to have only
  one way to do something. Plus it is unpleasant to have
  constructors that throw exceptions.

2000-07-22 Murray Cumming <[e-mail]>
* Removed custom.hh stuff - they are very complex
  preprocessor macros which are built from a perl script.
  This is unmaintainable anti-C++ hackery. If anybody
  wants it then they will have to use the MySql++ branch.
* Recombined the header files. Each file had 3 or 4
  headers, just so that almost every method could be
  inlined. Un-inlined lots of methods to make this
  possible. I did not see any need for most of them
  to be inline anyway. This is a classic case of
  optimising before you know what needs to be optimised,
  at the expense of structure and clarity.
* Put most classes in their own .hh and ..cc files,
  instead of declaring/defining several classes in each
  file.
* Every class has a constructor. This should prevent
  automatic copy constructors. There's no bitwise
  copying in C++!
* Every class has a destructor. Noticed that some
  classes have virtual methods but no destructor,
  though the destructor should be virtual.
* Removed exceptions-disabling boolean. Why have 2
  architectures?
* Removed const_string class. Used string instead.
  Removed const_string method overloads.

2000-07-15 Murray Cumming <[e-mail]>
* Put all classes and types into namespace mysqlcppapi
  Therefore removed define_short and undef_short files,
  and removed all Mysql prefixes. Now each type has only
  one name.
* Changed all structs to classes with default access of
  public. There's a lot of public member data that needs
  get/set methods.
* Added base class for exceptions: ex_base, which
  inherits from std::exception. Prefixed exceptions with
  'ex'.
* Put ex_BadConversion in it's own .hh and .cc file.
  Separated ex_BadQuery into .hh and .cc files. Gave both
  copy constructors and operator=. Did same for
  ex_BadNullException. Put them in their own sub folder.
* Put windows #includes into new windows_includes.hh file,
  instead of repeating them in almost every file.
* Removed connection0.hh - it's not used anywhere.
  
2000-07-13 Murray Cumming <[e-mail]>
* Changes at start of branch from MySql++ 1.7
  - Changed name to 'MySQL C++ API' (mysqlcppapi)
  - Changed to v1.7.1
  - Changed sqlplus directory to mysqlcppapi
  - Change sqlplus.hh to mysqlcppapi.hh
  - Made --with-mysql configure argument mandatory and
    improved explanatory text.
  - Removed <mysql++> header
  - All headers are now in mysqlcppapi directory.
    e.g. #include "mysqlcppapi/mysqlcppapi.hh".
  - Moved doc generation scripts from sqlplus directory
    to docgen directory - How do these work
  - Added mysqlcppapi-config script and AM_PATH_MYSQLCPPAPI macro
  - Added bad_query.hh to list of headers so that it installs.
  - Examples now need to be built by cd to examples and
    typing make. This allows for simpler project files.
  - Used GENERIC_* variables in configure.in instead of .version
    file. Now has release version indicating level of changes
    in interface.
  - Added explanatory comments to project files.
  - Changed licence to definitely be LGPL instead of grudgingly
    LGPL.
  - Removed some files from sqlplus directory, because they do not
    seem to be used: down.gif icon1.gif, icon2.gif, logo.gif
  - Moved some other files from sqlplus directory to a docgen
    directory in docs directory, because they look like they might
    be something to do with documentation, but I don't know.
    Hopefully somebody knows how the document generation works.
  - Added Makefile.am files to all folders to make sure that
    'make dist' works and includes everything.
  - Removed config.cygwin and makefile.win because they are
    obviously incorrect now. If anybody needs them then they
    can tell us what they are for and how to re-generate them.

Everything above this line refers to the 'MySQL C++ API' branch,
started from MySql++ 1.7

1.7 (May17 2000) Sinisa Milivojevic <[e-mail]>
                                       
* This is mainly a release dealing with bug fixes, consistency
  improvements and easier configure on some platforms
* A bug in fetch_row() method of ResUse class has been fixed. Beside
  changes that existed in a distributed patch, some additional error
  checking has been introduced
* A bug in escape manipulator has been fixed that could cause an
  error if all characters had to be escaped
* An inconsistency in column indexing has been fixed. Before this
  version, column names in row indexing with strings, i.e.
  row[<string>] , has been case sensitive, which was inconsistent
  with MySQL server handling of column names
* An inconsistency in conversion from strings to integers or floats
  has been fixed. In prior version a space found in data would cause
  a BadConversion exception. This has been fixed, but 100%
  consistency with MySQL server has not been targeted, so that other
  non-numeric characters in data will still cause BadConversion
  exception or error. As this API is used in applications, users
  should provide feedback if full compatibility with MySQL server is
  desired, in which case BadConversion exception or error would be
  abolished in some of future versions
* A new method in ColData class has been introduced. is_null()
  method returns a boolean to denote if a column in a row is NULL.
  Finally, as of this release, testing for NULL values is possible.
  Those are columns with empty strings for which is_null() returns
  true.
* Some SPARC Solaris installations had C++ exception problems with
  g++ 2.95.2 This was a bug that was fixed in GNU gcc, as from
  release 2.95 19990728. This version was thoroughly tested and is
  fully functional on SPARC Solaris 2.6 with the above version of
  gcc.
* A 'virtual destructor ' warning for Result class has been fixed
* Several new functions for STL strings have been added. Those
  functions (see string_util.hh) add some of the functionality
  missing in existing STL libraries
     * Conversion for 64 bit integers on FreeBSD systems has been added.
  On those systems _FIX_FOR_BSD_ should be defined in CXXFLAGS prior
  to configuring. Complete conversion to the usage of functions for
  integer conversion found in mysqlclient library is planned for one
  of the next releases
* A completely new, fully dynamic, dramatic and fully mutable result
  set has been designed and will be implemented in some of 2.x
  releases
* Several smaller fixes and improvements, including defaulting
  exceptions to true, instead of false, as of this version
* An up-to-date and complete Postscript version of documentation is
  included in this distribution
* Large chunks of this manual are changed, as well as README and
  TODO files.
  
1.6 (Feb 3 2000) Sinisa Milivojevic <[e-mail]>
    
* This is a major release as it includes new features and major
  rewrites
* Automatic quoting and escaping with streams. It works
  automatically , depending on the column type. It will work with <<
  on all ostream derived types. it is paricularly handy with query
  objects and strstreams. Automatic quoting and escaping on cout,
  cerr and clog stream objects is intentionally left out, as quoting
  / escaping on those stream objects is not necessary. This feature
  can be turned of by setting global boolean dont_quote_auto to
  true.
* Made some major changes in code, so that now execute method should
  be used only with SSQL and template queries, while for all other
  query execution of UPDATE's, INSERT's, DELETE's, new method exec()
       should be used. It is also faster.
* New method get_string is inroduced for easier handling / casting
  ColData into C++ strings.
* Major rewrite of entire code, which led to it's reduction and
       speed improvement. This also led to removal of several source
  files.
* Handling of binary data is introduced. No application program
  changes are required. One of new example programs demonstrates
  handling of binary data
* Three new example programs have been written and thoroughly
  tested. Their intention is to solve some problems addressed by
  MySQL users.
* Thorough changes is Makefile system has been made
* Better configuration scripts are written, thanks to D.Hawkins
  <[e-mail]>
* Added several bug fixes
* Changed Manual and Changelog
  
1.5 (Dec 1 1999) Sinisa Milivojevic <[e-mail]>
         
* Fixed bug in template queries, introduced in 1.4 (!)
* Fixed connect bug
* Fixed several bug in type_info classes
* Added additional robustness in classes
* Added additional methods for SQL type info
* Changed Changelog and README
  
1.4 (Nov 25 1999) Sinisa Milivojevic <[e-mail]>
         
* Fixed bug in store and storein methods
* Fixed one serious memory leak
* Fixed a very serious bug generated by gcc 2.95.xx !!
* Added robustness in classes, so that e.g. same query and row
  objects can be re-used
* Changed sinisa_ex example to reflect and demonstrate this
  stability
* Changed Changelog and README
* Few other bug fixes and small improvements and speed-ups
  
1.3 (Nov 10 1999) Sinisa Milivojevic <[e-mail]>
    
* Fixed several erronous definitions
* Further changed source to be 2.95.2 compatible
* Expunged unused statements, especially dubious ones, like use of
  pointer_tracker
* Corrected bug in example file fieldinf1
* Finally fixed mysql_init in Connection constructor, which provided
  much greater stability !
* Added read and get options, so that clients, like mysqlgui can use
  it
* Changed Changelog and README
* Many other bug fixes.

1999-10-15 Sinisa Milivojevic <[e-mail]>
           
* Changed version to 1.2					 
* First offical release. Version 1.0 and 1.1 were releases by Sinisa
  before I (Kevin Atkinson) made him the offical maintainer,
* Many manual fixes.
* Changed README and Changelog
* Changed source to be compilable by gcc 2.95.xx, tribute to Kevin
  Atkinson <[e-mail]>
* Added methods in Connection class which are necessary for
  fullfilling administrative functions with MySQL
* Added many bug fixes in code pertaining to missing class
  initializers , as notified by Michael Rendell <[e-mail]>
* Sinisa Milivojevic <[e-mail]> is now the offical
  maintainer.


1999-08-02  Sinisa Milivojevic <[e-mail]>

* Changed version to 1.1
* Added several bug fixes
* Fixed memory leak problems and variables overlapping problems.
* Added automake and autoconf support by [e-mail]
* Added Makefile for manual
* Added support for cygwin
* Added example sinisa_ex (let modesty prevail) which used to crash a 
  lot when memory allocation, memory leak and overlap problems were
  present. Smooth running of this example proves that all those bugs are
	fixed
* Changed manual
* Cheanged README
* Many other smaller things

1999-06-09  Michael Widenius  <[e-mail]>

* Changed version to 1.0
* Added patches from Orion Poplawski <[e-mail]> to support the
  UnixWare 7.0 compiler