[rescue] Total corporate madness (
Joshua D. Boyd
jdboyd at celestrion.celestrion.net
Thu Aug 7 16:18:23 CDT 2003
On Thu, Aug 07, 2003 at 04:09:10PM -0500, Joshua D. Boyd wrote:
> No, I just can't figure out how to make changes in the overly complex
> makefiles, and I can't figure out how to change configure script. If I
> had time, I'd rip out autoconf all together and make handmade files
> since this is an internal tool anyway. As it is, I've been removing
> reliance on #defines built dynamically by autoconf.
Oh, BTW, the GPL package I started with, I tried to send fixes and
features to the maintainers, but they didn't show any interest in them.
So, I figured I might as well make my life easier by chucking the parts
that were getting in the way, like autoconf, gettext localization, the
ability to work with special Microsoft binary image files, the online
help system (since I wasn't maintaining it anyway), and simplifying some
needlessly complex plans. I'm tempted to chuck support for some parts
that we don't use, but then again, we might someday use them, so support
stays for now.
Long term here, we would like to generalize this a bit more. Like, make
it support automated testing of boards (right now all testing must be
done very slowly by hand), a GUI, and personally, I'd like to see some
of the part specific code be moved to a scripting language, so that
adding support for a new flash chip or CPU doesn't require rebuilding
the whole program. The program supports grabbing information about the
chips boundary register from a text file, but it requires C code to say
what Hitachi or Arm or whatever pins to wiggle to program a flash chip.
That's just rediculus. Each CPU C file looks the same except for
certain char * constants. But, that part works fine (is just
unmaintainable), so it stays that way until after the project.
More information about the rescue