On 2024-10-30 08:38, Amos Jeffries wrote:
On 31/10/24 00:09, Francesco Chemolli wrote:
Hi all,
do we have any opinion on
https://www.ietf.org/archive/id/draft-ietf-httpbis-cache-groups-02.html
?
This is a further extension of Mark's variant Key header extensions to
HTTP.
When I proposed adding support for the first steps of these extensions
Alex complained that Squid's storage needed major overhaul before it was
worth doing.
I hope my assessment (referred to as "complaint" in the above paragraph)
was correct.
FYI; The "group" identifier in this HTTPbis Draft is akin to StoreID for
a "Vary marker" store object.
To support this mechanism cleanly Squid should first refactor "Vary
marker" to to have multiple headers determining further lookups objects,
not just the one Vary pattern.
FWIW, I doubt that group invalidation should be implemented using
something similar to Vary marker objects that Squid uses today. I
suspect there are better ways to instantly invalidate large sets of
cached entries. I assume that we do not need to debate the design (and
its prerequisites) now, but if somebody decides to actually proceed with
this project, they should keep this squid-dev thread in mind.
That is similar matching logic needed for the prior Key (also Vary-Key)
header mechanism which Alex has already vetoed as "too complex" to
implement.
Implementation complexity is an important factor, of course, but I doubt
that I have vetoed any mechanisms just because they are "too complex to
implement".
Alex.
_______________________________________________
squid-dev mailing list
squid-dev@lists.squid-cache.org
https://lists.squid-cache.org/listinfo/squid-dev
