[Tools-discuss] Re: Agenda Data Loading Slowly

[Rob Sayre <[e-mail]>]

On Sat, Oct 18, 2025 at 6:26 AM John C Klensin <[e-mail]> wrote:

>
>
> --On Friday, October 17, 2025 15:56 -0700 Rob Sayre
> <[e-mail]> wrote:
>
> > On 10/13/25 5:12 PM, Eric Rescorla wrote:
> >> So, it seems like maybe just remove the check for meeting number?
> >
> > Hi,
> >
> > It seems like there's too much detail on the older meetings. It's
> > fine to make me click through to another page to get the details of
> > the TLS WG meetings during IETF 122.
> >
> > Additionally, having been through this more than once, my advice
> > would be to look at the highest-volume read traffic and optimize
> > that. It can be something as radical as rewriting those in Go or
> > Rust, or just working on the cache. You can keep some control panel
> > page in really slow (but correct!) Python or Ruby forever.
> >
> > thanks,
> > Rob
>
> Rob,
>
> The counter-argument to a proposal like this is that, for some of us
> and especially for low-frequency participants, the information we are
> most likely to be looking for in poking through old agendas is
> information about WGs or work items we (and probably others) don't
> regularly follow.
>

Right, this is a normal social media problem. Let's look at this:

https://datatracker.ietf.org/meeting/119/proceedings

That is extremely detailed, and might make sense if you were looking at the
latest meeting, or heading for the next one. Just to pick some: who needs
to check in on all of IETF 119's meetings on "YANG model for management of
Network Tester", "Cross-Device Flow - Update IETF 119 (March 2024)", and
"Avoid Signaling RPKI Status in BGP - BCP Proposal".

That database query must be brutal. The old ones should never be generated
on the fly (it could produce a page every 5 minutes and let Nginx or
Cloudflare host a static page).

The old ones should also not have a page with every single meeting detail
and draft on them. That is because the value of timeliness in social media
(such as IETF minutes, not just Tweets or something) decays over time.

thanks,
Rob