> >/ Stop trading USB performance for Zylin performance, stop changing the JTAG > />/ API, improving the implementation is nice though. > />/ Listen to Laurent this time, he is right !! > / > This is a red herring. I do *NOT* intend to trade USB performance for > zylin performance. > > I'm doing this for a couple of reasons: > > - remove thousands of lines of code > - improve legibility > - yes - it improves performance of embedded hosts. This is not a bad > thing as long as I make sure that I don't deteriorate performance of > other interfaces when I'm done. > > Hang in there. If you give me a bit of breathing space, I'll see if I can't > get the worst of this done with in the next couple of hours.
DO NOT CONTINUE IN THIS WAY ! I sure/know you are trading USB performance, the old/actual performance *and* the future performance of *any* generic JTAG interface! The JTAG API with " in_handler _mask concept" has a BIG advantage to allow future JTAG Stream over USB and/or ethernet JTag interface and/or any serial link, with the advantage to keep the OpenOCD on the host computer. The JTAG API with " in_handler _mask concept" works for a long time. There are no reason to change it. The JTAG API with " in_handler _mask concept" was defined to be powerful and expendable as *first* Priority. The JTAG API with " in_handler _mask concept" was not defined to improve legibility as first Priority -> performance is much important than legibility ! If the JTAG API with " in_handler _mask concept" do not match your hardware, change the hardware but do not change the JTAG API for a specific hardware ! PLEASE RESPECT THE GENERAL CONCEPT OF OPENOCD ! PLEASE MAKE SURE TO COME BACK WITH "in_handler" and "_mask" before the end of 08-MAY-2009 ! Regards, Laurent http://www.amontec.com _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
