[OLPC-devel] Re: Minutes from System Software telecon, 2006-08-22

Richard Smith smithbone at gmail.com
Wed Aug 23 10:47:35 EDT 2006


> > >nevermind :)). Is that supposed to happen?
> >
> > Doesn't happen for me.
> It should - if it doesn't, then you did something wrong.

I misuderstood.  The blank is while kbl-kexec loads, yes.

> > - Buildrom dependency info for olpcflash.c doesn't work right.
>
> By design - buildrom works on a macro level, not a individual file level.
> It uses the existance of the final binary to determine if it needs to rebuild
> or not, not the individual .c files.  Thats why the olpcflash-clean target
> exists.

Ok.  That makes sense.  I thought I saw the linuxbios part of that
attempts to detect when you have make a change to the config file and
try to rebuild.  I extrapolated that the other targets would do the
same.

So is the following is true?

When do a get pull you either need to clean and rebuild the entire
tree or you have to look at the diffstat and make pkg-clean on every
package that has changed in the tree.

If so then we probally need to document it somewhere and have a reminder.

> No - bootmenu needs 'ESC' to pop up the boot screen.  The keypad worked by
> luck because if numlock is off, they issue ESC commands which are enough
> to bring up the menu.  I'm not sure where you got the numbers thing.

I got the numbers because the previous loader we had before bootmenu
took numbers to select the boot option and  the linuxrc script checks
for return codes  0 1 2 3  or  -1.  I made a bad assumption that
bootmenu would do the same.   When presing the number keys didn't work
I started looking as to why.
I saw the ESC in the code and knowing that the keypad does an escape I
jumped to 123 on the keypad.  Since it just "worked" I didn't really
look at the code hard enough to see I was wrong.  My bad that I didn't
verify my scan codes.  Thanks for the correction.

> enter to select it.  Anything else that works is just plain luck.  The source
> code is quite clear on this.

At 3:30am, I admit,  my clarity was suffering.  I probably should have
quit before then.  So cool.  None of this is an issue then.

-- 
Richard A. Smith



More information about the Devel mailing list