Does anyone on the list have tips for reducing compile times for MySQL++
code (read: template instantiations) with g++? I have a source file that
contains 6 ResUse and 6 Row objects and it takes g++ 4.0.3 over 2
minutes to compile it on my 1.8 GHz/512 MB machine.

I've used the -frepo option in the past, but because most of my source
files use both MySQL++ code and Boost code, I cannot use the -frepo
option (Boost breaks with -frepo).

When I search Google, I find many references to -fexternal-templates,
but this option is deprecated in g++ 4.0.x. However I can't find a good
explanation or discussion of why it has been deprecated or what should
be used in its stead.

My code uses 1 Connection object and 1 Query object. It uses 1 ResUse
object and 1 Row object per call to Query::use(). I'm guessing that I
could reduce compile time by instantiating fewer ResUse and Row objects.
I've read numerous posts about elusive bugs due to reusing ResUse and
Row objects. Here's a generalized form of my code:

Connection con;
con.connect(...);

try {
Query qry = con.query();
qry << "select ... ";
ResUse res1 = qry.use();

try {
while (Row row = res1.fetch_row()) {
...
}
} catch (const EndOfResults& lastrow) {
// Do nothing
}

qry.reset();
qry << "select ...";
ResUse res2 = qry.use();

try {
while (Row row = res2.fetch_row()) {
...
}
} catch (const EndOfResults& lastrow) {
// Do nothing
}
} catch (const BadQuery& badq) {
// crash
}

Drew Vogel



--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw

Drew Vogel wrote:
I take the view that C++, STL and MySQL++ do a tremendous amount of work
for the programmer, and they do most of that work at compile time. In
exchange for this, you get a program that runs very quickly -- compared
to, say, a Perl DBI based program -- and doesn't sap months of
development time.

If you wanted to trade development time for compile time, you could use
the MySQL C API instead.

I don't mean suggest that you take the option. It's just that there are
no free lunches, and you should know what you are trading to get the
benefits you have.

--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw

On Tue, Jan 17, 2006 at 11:53:41AM -0600, Drew Vogel wrote:
I haven't found mysql++ to cause super long compile times (perhaps I'm not
using it strenuously enough ), but I have found certain parts of Boost
to do it.

ResUse and Row don't seem that complicated to cause such a long compile time.

- Chris


--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw