The basic concept of MySQL++ v2.0 is to roll up all of the ABI breakages
that we've been putting off. Ideally, this will be the last ABI
breakage for a long while. The mad god Murphy may have other plans, but
even if we have to change the ABI again in the next major revision to
correct unpopular changes, it's still in our best interest not to put
off these changes any further. The earlier we discover problems with
these changes, the less time until we settle on a stable ABI.
We're therefore putting off major changes that shouldn't impact the ABI
to v2.1 or later. See the current Wishlist in the Subversion repository
if you want a list of changes planned for future versions.
This plan is not engraved in stone. Some of these changes may get
pushed to v2.1, and other items from the Wishlist might get rolled up
into v2.0.
The planned v2.0 changes are:
o Create interface class for disabling exceptions, and derive all
classes that currently can have exceptions disabled from this
interface. Then create an exception-disabling class that
is designed to be created on the stack with a reference to
one of these objects; while the disabler object is in scope,
exceptions are disabled on the target object. Side benefit:
this will standardize the mechanism for temporarily disabling
exceptions, which currently varies somewhat between classes.
Also possible to do this with templates, using compile-time
polymorphism.
o Move SQLQueryNEParms into exceptions.h.
o Hoist all common exception functionality up into a superclass,
and derive all MySQL++ exceptions from that.
o Extract Connection class's lock mechanism into a separate class,
instances of which could be created on the stack so the lock
is always relesased on function exit. Derive Connection from
an interface class ("Lockable"?) that provides lock and unlock
functions. Add a configure script option to allow this lock
mechanism to use platform mutexes via the Boost.Threads library.
Also derive Query from this new interface, as it also has a
locking mechanism.
o Change Connection::create_db() and drop_db() to return true on
success.
o Connection::create_db(), drop_db() and select_db() should take
their arguments by reference, and be const.
o Remove Connection::infoo() (Apparently an attempt to cope with
typos for "info()".)
o Rename Connection::read_options() to set_option().
o Rename Connection::real_connect() to connect(), drop old
connect() function, and set defaults for all parameters
beyond what connect() supported. Thus the API only changes
for real_connect() users, and it's just a function rename.
o Make Connection::execute() use mysql_real_query() instead of
mysql_query() so it can handle binary data.
o Remove query-building interface from Connection. It just
parallels most of the Query class interface, for little benefit.
o Fold SQLQuery class into Query.
o Use mysql_real_escape_string() instead of mysql_escape_string()
within the Query::pprepare() function. (In C API
since 3.23.14.) Depends on previous change, because
mysql_real_escape_string() requires access to the current
Connection object, to get the database's default character
encoding.
o See if we can use mysql_real_escape_string() in manip.cpp
as well. (This might end up being pushed into v2.1, if it's
too tough.)
o Give combined Query+SQLQuery class all the members it needs to
look like a std::stringstream. Use a private stringstream
data member to hold the query-under-construction.
o There are comments in sql_query.h saying "only here temporary".
They've been there for literally years. Time to decide what
to do about them!
o Set type to some default in ColData_Tmpl default ctor.
o Fold RowTemplate class into Row.
o Collapse mysql_* date and time classes into their subclasses,
Date, Time, and DateTime.
o Make dtors virtual in classes with virtual functions, to squish
GCC 4.0 warning. Also, go through each changed class looking
for any code that allocates memory or other resources, and
make sure they're released in the dtor.
o Collapse multiple public and private sections of class
mysql_type_info into one of each. Also, consider renaming
that class.
o Collapse store(string&) and store(string&, bool) into a single
function by making bool argument default to false. Same for
use() and execute().
o Several of the value_list* and related functions in vallist.h
and row.h can be collapsed into one by use of default
parameters. A conservative refactoring wouldn't change the
API, but keep an open mind about whether a more aggressive
change would be better in the end.
o Try to collapse all SQLQueryParams::set() overloads into
a single member by suitable use of default parameters.
o Remove simp_list_b from vallist.h. It's not used within the
library, and it doesn't seem to be useful for end-user code.
o Null's copy ctor should take its parameter by reference.
o Remove the field name form of field_list(). It's pretty silly:
it takes a field name as a paramter, and returns a field name!
True, there's a type conversion going on here, but...
o Change Row ctor to take the MYSQL_ROW by const reference.
o Swap the definitions of Row::operator[] and lookup_by_name().
That is, have the operator take a field name, and define a
function (at()?) that takes a field index. The original problem
solved in 1.7.10 still exists, in that modern compilers won't
let you overload the operator on both strings and integers,
but the string form probably is the most useful.
o Decide on a new soname scheme? It'd be nice if the soname had
some obvious relationship to the library version. We could
actually go with a 1:1 relationship, and simply override the
default library symlink scheme to handle things correctly if
libraries from the old scheme are still present.
o Add a userman section severely summarizing ChangeLog, with a
focus on API and ABI breakages.
o Automatically define USE_WIN32_UCS2 in util.cpp when platform.h
detects a Windows compiler. Test with VC++ and Cygwin in
-mno-cygwin mode.
o Remove the exception handling stuff from all the simple*
examples, so they earn their name. Newbies get thrown into the
deep end of the pool in simple1.cpp in the userman right now...
o With previous change done, promote existing "notes on
exceptions" section of tutorial into a new section, or even
a new chapter.
o Consider splitting "The Basics" section of tutorial into two
parts, with the new EH discussion separating discussion of
the examples that use EH and those that don't.
o Add thread support to examples/Makefile.simple. Commented-out
by default, but available if required.
o It may be possible to optimize the use of ColData in the return
from Row::operator[]. Currently, that operator returns
a temporary ColData object, which contains a std::string
buffer which is initialized by a const char* pointer to
data within the Row object. Since the ColData object is
temporary, you currently must copy the data a second time to
store it when using Row::operator[]. If the end user just
wants a const char*, this double copy could be prevented.
See http://lists.mysql.com/plusplus/4451 for the proposal.
o Relicense user manual. Linux Documentation Project License v2.0
is the current front-runner.
