- From: Daniel DuBois <ddubois@spyglass.com>
- Date: Tue, 09 Apr 1996 22:03:16 -0500
- To: www-logging@w3.org
At 02:38 PM 3/27/96 -0500, hallam@w3.org wrote:
>#Start-Date: 1996-01-12 23:00:00
>#End-Date: 1996-01-13 21:00:00
>#Date: 1996-01-12
>23:20:00 GET /foo
>23:30:00 GET /foo
>#End-Date: 1996-01-13 21:00:00
>
>Note this is optional... The semantics for the start and end date are global,
>they apply to the whole file and are there purely so that a person can come
>along find out what the contents of a log file are using more on the head.
>The date fields are positional however.
I've thought about this some more, and I have these questions.
Are you saying the End-Date and Start-Date directives are *strictly* global,
and that the situation I described above with starting and stopping are
inappropriate/illegal? If so, I still disagree.
When Joe Webmaster shuts down the server, the server will, as a matter of
course, append a "#End-Date:" directive on that log file. If he starts back
up, with the same file name, we naturally go into append mode. We certainly
wouldn't over write the old log file, and attempting to go back and
automatically edit out that existing "#End-Date:" is not a viable option.
Therefore, I think for a stopped and restarted server, a pair of End-Date
and Start-Date directives embedded in the middle of the log is appropriate,
if not a requirement.
Besides, what if Joe logs to foo.txt for ahile, then bar.txt, then back to
foo.txt. It will be alot clearer what was occuring than trying to decipher
Date directives and attempting to calculate areas of non-overlapping time
values.
-----
the Programmer formerly known as Dan
http://www.spyglass.com/~ddubois/
Received on Thursday, 11 April 1996 11:05:20 UTC