myexperiment-hackers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [myexperiment-hackers] Problem with /home caching


From: Jiten Bhagat
Subject: Re: [myexperiment-hackers] Problem with /home caching
Date: Fri, 17 Oct 2008 10:48:23 +0100
User-agent: Thunderbird 2.0.0.17 (Windows/20080914)

We discussed this in yesterday's chat (see logs) and came to the conclusion that a joint mechanism that uses Tom's caching stuff AND a 1 hour expiry timed cache would be a good option. What do you think?

Dave N also mentioned the client side updating of time info which could potentially work too.

All these require scheduling and resourcing according to THE PLAN :-)

I guess some of this will change with Sergey's news code.

Sergey's news code just affects the "my news" box on the home page (and other user and group show pages) and will be the mechanism for subscriptions that require events spanning *different* models. It won't be used for the other boxes on the home page, as those are events pivoted on one model and using the ActivityLog stuff would mean having to at least one extra query, with extra logic, and also require it's own caching.

Jits


Danius Michaelides wrote:
On Thu, 16 Oct 2008, Jiten Bhagat wrote:

Hi all

See attached screenshot....

Tom's caching work has improved performance of the "Home" page substantially, but there is an issue we need to fix - the time information (eg: "6 hours ago") gets stale and can cause confusion (I've now seen that "6 hours ago" message two days in a row) and won't get updated till the fragment for that box gets invalidated (by someone uploading or editing something).

Any ideas?

a) dont cache
b) dont write the time info
c) calculate the time info on the client (although I guess the list would
   still get out of date)
d) make the list dynamic (like annoucements on the frontpage)
e) make the cache observers invalidate the correct user caches

I guess some of this will change with Sergey's news code. I'll make
option (a) live in the meantime. Having the improved auth code would also
help here.

PS: this is why I had suggested a while back that timed expiry caching was the least disruptive option.


It'd still give incorrect text to the user. If you recall the timed expiry
was a temporary solution until it was done properly.

Danius


_______________________________________________
myexperiment-hackers mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/myexperiment-hackers





reply via email to

[Prev in Thread] Current Thread [Next in Thread]