Page MenuHomePhabricator

Massviews Analysis with External link as source produces same result set, irrespective of the external link
Closed, ResolvedPublicBUG REPORT

Description

I'm using Massview Analysis https://tools.wmflabs.org/massviews/ with External link as the source. (from this table, all the links in the 3rd column)

For the external link I choose http://www.kb.nl, with nl.wikipedia.org as the platform. This gives the URL

https://tools.wmflabs.org/massviews/?platform=all-access&agent=user&source=external-link&start=2018-02-21&end=2020-02-05&project=nl.wikipedia.org&sort=views&direction=1&view=list&target=http://www.kb.nl

which outputs a list of 141 results

Knipsel.PNG (783×1 px, 453 KB)

So far so good....

but when I return to the form and enter another query, lets say for http://www.delpher.nl, so via https://tools.wmflabs.org/massviews/?platform=all-access&agent=user&source=external-link&start=2018-02-21&end=2020-02-05&project=nl.wikipedia.org&sort=views&direction=1&view=list&target=http://www.delpher.nl

I get the exact same result page as for www.kb.nl (so same screenshot as above). It looks as if the new url (http://www.delpher.nl) is totally ignored. I observe that it seems to be serving the result page from cache, rather than actively building a new result page for the new url

I'm working with Firefox on Windows 10, and also tried using Chrome on Windows 10, but that gives me the same result. I also tried erasing browser cookies etc. but to no result as well.

Can anyone reproduce this, and point out where the problem might come from and what solutions there might be?

Also described on https://meta.wikimedia.org/wiki/Talk:Pageviews_Analysis#Massviews_Analysis_with_External_link_as_source_produces_same_result_set%2C_irrespective_of_the_external_link

Event Timeline

MusikAnimal triaged this task as High priority.

Fixing now!

MusikAnimal moved this task from Backlog to Done on the Tool-Pageviews board.

Fixed! Sorry about that. I think this broke during a recent refactor.

I also fixed another bug where it wasn't iterating through all the external link results properly. So for your first example, there were actually 195 results, not 141. That bug I think has been around since the beginning :(

Let me know if anything looks off, and thanks for filing this report!

Excellent, seems to work very well again. Did some testing for multiple/different URLs, all seems well, no more weird "caching".

As far as the updated numbers are concerned: I did some reconstructions of results I measured 2 years ago (this table): the reconstructed numbers are pretty close, so this seems to be OK as well