Talk:Analytics/Archive/Data/Pagecounts-all-sites

Rendered with Parsoid
From Wikitech
Latest comment: 9 years ago by QChris in topic No file generation since 2014-10-10

While it would be nice if the files themselves contained a reference to this documentation, the current format of the files do not allow comments.

Adding a convention like “Every line that starts in # is a comment” would be easy, but it would probably break scripts of people.

Is having the files themselves link to this documentation worth it? qchris (talk) 11:41, 2 October 2014 (UTC)Reply

Oppose (weak) Breaking backwards compatibility to pagecounts-raw and potentially breaking scripts of others eat up the benefits for me. qchris (talk) 12:05, 2 October 2014 (UTC)Reply

Abbreviation for mobile site

For now we went ahead with not treating the mobile site special. This buys a nice, and simple abbreviation mechanism, without unnecessary ifs.

For example, regardless of the chosen site, when looking for the pageviews of Berlin on wikivoyage, the abbreviation mechanism is always to replace “.wikivoyage.org” with “.voy”.

  • en.wikivoyage.org -> en.voy
  • en.m.wikivoyage.org -> en.m.voy
  • en.zero.wikivoyage.org -> en.zero.voy

However, eight .wikimedia.org wikis get abbreviated to .m too. This does not break the 1:1 correspondence between abbreviated domains and domains. But it's not as smooth as one wants it.

But alternatives—like using a special abbreviation for the mobile site on wikipedia (E.g.: en.mobile for en.m.wikipedia.org)—would make the use of .m a bit clearer, but would cause more ifs for people that want to consume the files.

Also, we want to stay backwards compatible with the pagecounts-raw files.

Should we keep the current use of .m for the mobile site? --qchris (talk) 12:59, 2 October 2014 (UTC)Reply

Support Support I'd keep things as is --qchris (talk) 13:01, 2 October 2014 (UTC)Reply

zero stats

First, congratulations on getting this up and running -- super-exciting. :)

The zero.* stats just capture the relatively small traffic that goes to zero.wikipedia.org, and not zero-rated traffic to m.wikipedia.org, correct (see traffic dashboard, where zero.wp.org was inflated for a while due to an automated script)? I'd like us to make this clear, if so, so people don't draw incorrect conclusions about the volume of traffic.--erik (talk) 20:42, 2 October 2014 (UTC)Reply

Yes Done Added the “Requests to the zero site and Wikipedia Zero requests” section. --qchris (talk) 13:48, 3 October 2014 (UTC)Reply

No file generation since 2014-10-10

Just after I wrote the code to switch from the (non-mobile, non-zero) pagecounts-raw, it seems this service hit a hiccup. The last available file is "pagecounts-20141010-150000.gz" as I write this, two days since the production of that file. I appreciate this service and hope the issue can be resolved quickly. Thanks, West.andrew.g (talk) 05:48, 13 October 2014 (UTC)Reply

I see things are now back up and running. Thanks so much! West.andrew.g (talk) 00:07, 14 October 2014 (UTC)Reply
Yes Done Thanks for reporting. Ja, we noticed yesterday that a service in the pipeline had been stopped for a deploy but has not been restarted afterwards (See bug 71994). All missing files have been recomputed. Sorry for the inconveniences. --qchris (talk) 07:21, 14 October 2014 (UTC)Reply