Re: [Passwordsafe-devel] Timestamps, character encodings in .dat files

[Philip N. <phi...@gm...>]

On Sat, 11 Dec 2004 14:48:38 -0500, Frank Pilhofer <fp...@fp...> wrote:
> An 31 bit integer (i.e., the positive range of a 32 bit 2's complement
> signed integer) gets you to the year 2038 only, that's right. But a 32
> bit (4 byte, unsigned) integer gets you to all the way to 2106.

Ah, right. The "unsigned" bit hadn't fully registered.

> The .dat file format does not have a use case for time stamps in
> the past. I don't think any of the passwords in current use were
> created before 1970.

OK, this makes sense, and also obviates the need for a signed type.

> > "network byte order" [...]
> 
> That was for consistency; .dat files use little endian integers all
> over the place, e.g., for field lengths.

OK, a good argument.

Cheers,
-- 
Philip Newton <phi...@gm...>