From: Bruce Thompson <bruce_at_otherother.com>
Date: Tue, 5 Nov 2002 11:35:43 -0800

On Tuesday, November 5, 2002, at 09:13 AM, Linus Nielsen Feltzing wrote:

> Quoting Bruce Thompson <bruce_at_otherother.com>:
>> As you mention there is potential for damage if a write fails
>> badly. It seems to me that spinning up the disk is the biggest power
>> drain and thus the most likely to fail if the batteries are low.
>
> My general approach to disk writes are: don't write to the disk if it
> isn't
> necessary. Updating the read access time falls into that category.

Fair enough, though if the JB6K knew the time then I might consider
creating a branch with that feature...

>> Perhaps the access time could be written right after the initial read
>> when starting a file.
>
> Of course. When would we do it otherwise?

One could argue that the last access time is the last time anything was
read from the file. That's not the direction I would consider though.

>
>> Another possibility could be to check the battery
>> level and if it's below a low watermark then don't do the write.
>
> We already do that, actually.

Excellent, I had hoped that would be the case.

>
>> This does beg the question: Does the JB6K know what time it is?
>
> No, it doesn't.

Damn. :-) So the whole thing is moot for me... Oh well.

>
> /Linus
>
Received on 2002-11-05