Help talk:Cite/en

Is the error message hidden in Page namespace?

The whole point of the "follow" functionality is to be used in Page namespace when the first named part of the reference is on a different page. Therefore, it would be disadvantageous to display an error message in Page namespace when the "follow" functionality is being used as designed. ~~~~

Further to that its whole purpose is based on the fact that the second part of a footnote/reference is separated and will be joined upon transclusion, so

1. the second part of a footnote ("follow" component) will not have its predecessor ("name" component)
2. the second part ("follow" component) of a footnote displays differently, usually without the ordinal number, so it is not an error, and it must not impact on the visual representation.
3. the joining together in an end note ("name" + "follow" components) displays all of the concatenation of name, follow1, follow2, ..., follown (best example I have seen is about 6 follow refs)

Examples of use in the wild

• s:en:Page:The life of Matthew Flinders.djvu/82 (ref name=)
• s:en:Page:The life of Matthew Flinders.djvu/83 (ref follow=) note display with lack of ordinal, your change would cause an error for first ref, and display second (presumably)
• s:en:The Life of Captain Matthew Flinders, R.N./Chapter 4 (references)

Please hold the roll-out. If the concept of its implementation and its use is not understood, and hasn't had primary and timely consultation with the appropriate communities, then a rollout seems premature.

Note: The deWS does not use ProofreadPage, and the primary usage of "follow" is in the Wikisources through the implementation of the ProofreadPage Extension. Where was the consultation with Wikisources?

When ThomasV put in the original adaption to Cite to have "follow" parameter, it was clearly annunciated at the time its reason, its purpose, and where it was being utilised. I remember the hullabaloo about getting any changes to ProofreadPage and Cite through that time (c. 2011???)

To tie in some of the history in case it is every needed

• s:en:Wikisource:Scriptorium/Archives/2011-03#Running examples of long references splitted into more than one page time of rollout
• Special:Code/MediaWiki/71157

though I cannot quickly find ye olde bugzilla link

And if I don't say so myself, "shoots and scores" on the timeline.

Looking at releng's procedures, it's too late to pull this from the ongoing deploy (the impact isn't severe enough to hold a deploy, and rollout to Group 0 wikis started yesterday). Which means the available options are to roll this change back through a SWAT deploy after the train hits Group 1 wikis, or in the first scheduled deployment next week (1.35.0-wmf.19, which hits enWS on Wed. 12 February).

A little update: Our team will try to either remove the change from the train or to remove the change after the train update has happened. We’re currently investigating the options and will let you know as soon as we know. Many thanks to everyone who pointed the issue out!

Note that if, as seems likely, you are not able to prevent it from being deployed, it would be useful for us to know that in advance so that we can prepare for it. Similarly, if we know the SWAT window for a rollback it's generally easier to communicate to the community with “Known issue, will be fixed <at time X>”.

Hi, an update from our side: we were able to apply the rollback just a moment ago so tonight's train deployment will not include any fun "breaking" changes. Thanks for the help and for explaining the feature to me, here and in Phabricator!

Big thanks to everyone involved for scrambling at the last minute here: much appreciated!