On Wed, Dec 17, 2008 at 12:01:05PM +0100, Øyvind Harboe wrote: > > *) The xscale target does something similar with its debug binary, which > > tells me it's acceptable if not preferable. > > I vote for keeping it simple & robust. The procedure should be > described and committed > to SVN, but it can be done manually for files which are updated once in a > blue moon.
It is described in the README file in the new rlink directory. If there is a better place to do that, it can certainly be changed. This is, indeed, one of those blue moon situations. People aren't going to be editing the DTC code (init.m4 and call.m4) much. > I'd suggest committing these files to SVN, but keeping them > out of the way for the ordinary citizen. A separate folder is nice. > Works well for > e.g. xscale. So does that mean that what I have is good? > > We've got lots of other examples, e.g. the at91fr40162 flash driver contains > a .elf file that is compiled with an arm-elf toolchain using eCos source. > We're *NOT* going to make that part normal OpenOCD builds! :-) I had a similar (though less radical than needing eCos source) concern with the requirement for perl and m4. Though those are both very common programs, it is probably best to not assume that everybody who's building OpenOCD will have them installed. Granted, one could assume m4 if one is running bootstrap, but perhaps not if they're just running ./configure in a distribution tarball. _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
