On Sun, Feb 24, 2002 at 10:31:36PM +0100, Jonathan Schmitt wrote:
> I think I have (with some delay) located a second one billion bug.
> When searching for the newest version of one file with a little script, I
> used the date command (with options -r,+%s).
> The result was not, as expected, the latest file, but the latest with a time
> less than one billion. This one isn't of course a native second one billion
> problem but more a problem of the bash-tests. As this hasn't been done on a
> debian system (but bash 2.05a). Perhaps anyone can verify this is also
> present on debian systems (I'm currently far far away from a running one)?
If I am following you correctly, this is probably a sorting problem.
Sort the times as ctimes in numerical, not alphabetical order. Since:
99999999 < 1000000000
"999999999" > "1000000000"
Where "" indicates a string. In perl (and I know you're using bash), the
difference is `sort ($a $b)` (default alphabetical) versus
`sort {$a <=> $b} ($a $b)` (numerical).
What exact command did you use?
Perhaps.
James
--
James Bromberger <james_AT_rcpt.to> www.james.rcpt.to
Remainder moved to http://www.james.rcpt.to/james/sig.html
Attachment:
pgp1ELT3BDEQ0.pgp
Description: PGP signature