any drawbacks to using copy-nand and save-nand to install XO images

[Mitch Bradley]

James Cameron wrote:
> On Thu, Mar 27, 2008 at 03:23:59PM -0400, Michael Stone wrote:
>   
>> A much better strategy is to reflash an XO, boot it off of external
>> media (like a USB key), make changes to the NAND, then save-nand, thus
>> avoiding the first-boot configuration junk.
>>     
>
> I agree, and this is the way I did the 406 recut with changed wireless
> firmware a few months ago.  Without doing it this way, the first-boot
> configuration changes could get in the way of normal operation.
>
>   
Food for thought:

I would argue that the existence of first-boot configuration changes is 
in itself a problem.

In principle, the OS shouldn't have to "fix itself" the first time it boots.

I realize that, due to realities of the way things work right now, 
changing that is easier said than done, but nevertheless I think it 
would be good to keep the principle in mind.

As one example, the "olpc-configure" script generates some state files 
(.olpc-configured) to report that it has already run and need not run 
again.  We have had lots of problems e.g. in manufacturing with that 
state becoming stale.  But the steps the script performs to derive that 
state are quick, so one could consider just regenerating it on every 
boot (in the absence of a user-supplied override of the auto-generated 
results), perhaps even putting it in a RAM FS to minimize artifacts in 
the persistent filesystem.

Any persistent state that you really don't need is just a problem 
waiting to happen.