KDevelop breakpoints: sending to gdb optimization

Vladimir Prus ghost at cs.msu.su
Fri Dec 2 10:49:04 UTC 2005

Hi John,
I have a question about breakpoints management in KDevelop.
At the moment, KDevelop tries hard to avoid sending a command to gdb if 
breakpoint has no changed.

There are two levels of checks. First, when QTable emits valueChanged signal, 
KDevelop checks that the value is indeed different from the value in 
breakpoint. Second, each modifier in the Breakpoint class compares the
new value with the old and sets a  flag if the new value is different. GDB 
controller uses those "changed" flags to decide if to send various command, 
like "condition", or "enable".

But, the time to send a single command seems very small. And those two levels 
of checks add quite some complexity to the code.

Do you see any hidden dangers in removing that optimization, and just sending 
all data about breakpoint to gdb when user touches any table cell?


More information about the KDevelop-devel mailing list