On 7/27/20 9:40 AM, Miroslav Suchý wrote:
Dne 16. 07. 20 v 21:56 Brendan Early napsal(a):
and would time out if an external service took too long to respond.
Only if you wanted to list builds, bugs.... the overview page is more or less static (i.e. it is rendered just from local DB).
I was basing that statement off of a quick skim of old fedora packages source. I may be mistaken as I'm still new to python, but I was under the impression that it always queried koji when it was given a request.
https://github.com/fedora-infra/fedora-packages/blob/master/fedoracommunity/...
Fedora-packages-static does not list them at all. This is ok when we agree that this is what people find sufficient.
This is intentional, I can easily write some client-side JS to add that functionality if it is needed.
https://pagure.io/fedora-packages-static/pull-request/1#comment-0
What I am curious about fedora-packages-static - how long it takes to generate all those pages? When I tried to retrieve initial data fedora-packages-ng, it took several hours. Which may be too much for daily job.
How are you getting the data in ng?! From a clean state on my computer, it takes packages-static around 140 seconds to get the package metadata into the db and about 195 seconds to generate all the static pages.
How much data it consumes?
Not anything monumental, just ~6.3G.
What is yacysearchserver-pkgs-playground.apps.os.fedorainfracloud.org?
A staging YaCy server that Timothée had on communishift.
https://lists.fedoraproject.org/archives/list/infrastructure@lists.fedorapro...
Or when we delegate search to Google, how long it takes to refresh the search index?
I have code to generate sitemaps, which can be used to tell Google how often we want to be crawled and it will make sure every page is indexed. This should improve discoverability, I don't know how well the old one was indexed.
Will be people ok with the speed?
I don't understand exactly what you are referring to here. If you mean the speed of search, I don't specifically know about the speed of Solr. In the event hosting something ourselves does not work out we could always go with Algolia, which is near-instant.