Duane Wessels wrote:
> if the response does not include a Last-Modified reply
> header, then entry->lastmod = -1.
Yes!
> We will NOT make IMS validations for entries with
> ->lastmod == -1.
Why not?
I agree that we use some threshold (like the MIN refresh pattern), but
after that we might just as well use IMS. Why bother to adding a special
case that we should not? Since there is no Last-Modified date the IMS
date should be the reply date.
> Instead the 'refresh_pattern' MIN parameter will be moved
> to the end of the refreshCheck() algorithm and it will apply
> to cache hits which lack ->lastmod. Thus, the MIN parameter
> no longer ALWAYS violates HTTP, but only comes into effect
> when there is no Expires: and no Last-Modified: reply header.
Close to what I have done. My current patch gives MIN a higher priority
than Last-Modified, but this was because entry->lastmod was set on every
object and I did not want to get into any magic to determine when it was
valid or not.
To please the tweaking ones, refresh_pattern should have options to let
MIN override both Last-Modified and Expires.
/Henrik
Received on Tue Jul 29 2003 - 13:15:51 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:11:49 MST