Re: Patch for new Timeouts

["Roy T. Fielding" <fi...@liege.ICS.UCI.EDU>]

I don't like this one, sorry.  You can only have one active alarm
at a time, and doing a string comparison to determine which alarm
should be set is very bad.  The only valid way to do multiple timeouts
is to maintain an alarm table, select which alarm(s) are being set
with a bitmask, and call alarm() with the lowest table entry which
is greater than zero.

......Roy

Re: Patch for new Timeouts

[Ed Korthof <ed...@organic.com>]

Not hard to do it the way you describe; but looking through the code we
have, there aren't that many hard_timeouts called, and certainly none
which would be mistaken by a string comparison.  Still, this makes sense
-- I'll submit a revised patch later today or tomorrow (also, done so that
KeepAliveTimeout can be folded in).

     -- Ed Korthof        |  Web Server Engineer --
     -- ed@organic.com    |  Organic Online, Inc --
     -- (415) 278-5676    |  Fax: (415) 284-6891 --

On Tue, 21 Jan 1997, Roy T. Fielding wrote:

> I don't like this one, sorry.  You can only have one active alarm
> at a time, and doing a string comparison to determine which alarm
> should be set is very bad.  The only valid way to do multiple timeouts
> is to maintain an alarm table, select which alarm(s) are being set
> with a bitmask, and call alarm() with the lowest table entry which
> is greater than zero.
> 
> ......Roy
>