[coyotos-dev] coyotos.sleep, sleepTill, the epoch

[Valerio Bellizzomi]

On 14/02/2007, at 14.04, Pierre THIERRY wrote:

>Scribit Valerio Bellizzomi dies 14/02/2007 hora 02:36:
>> > For both of these purposes, capturing the time of first boot is
>> > equally good.
>> If for you it is equally good, we can live with first boot time.  I
>> give up, however it isn't equally satisfactory for my purpose.
>> Probably I will send later a detailed note of why it isn't.
>
>I must admit on don't see how the first boot time wouldn't be
>satisfactory for forensic purpose. The only difference that installation
>time could do is if different versions of the installer would me used at
>different times, but then, installation time would not be a good
>indicator at all.

There is no real problem in forensics, the usefulness is in diagnostics.

> Would a combination of an identifier of the installer
>and first boot time be satisfactory, then?

This is a very good idea. I think a record of both install date and
installer identifier would be the best. Here is why:

It occurred to me that the first boot can *fail*. The reasons could be a
syntax error or bad system specification in the mkimage input file.
When this occurs, it is very useful to have such record for reference in
debugging of mkimage input file, and test report writing.

Note that we test our disks before installation, but the diagnostics are
never 100% reliable. It can happen that we install coyotos on a tested
disk that fails on boot. In this case the record is not useful for
debugging, but it might still be useful to know when the disk was last
written (if the data is not corrupted).


val