Page MenuHomePhabricator

Convert selected external links to HTTPS
Open, LowPublicFeature

Description

From a recent WP:VPT discussion, it was suggested to have the following added as an automatic "general fix" in AWB: changing the protocol for certain external links from HTTP to HTTPS, although only within citation templates. For now, this should include a number of Google services (e.g. Google Books, Google News, YouTube) and the Internet Archive (i.e. Wayback Machine). In essence, this should probably be some sort of RegEx feature that looks for |url=http://books.google.[com|co.uk|ca|de|it|es|...] and replaces it with |url=https://books.google.com (yes, the top-level domain switch is intentional here, since Google redirects from .com to a specific country version, and we should keep the links general). The same for Google News, YouTube as well as http://archive.org and http://web.archive.org/ links. (For the latter, of course, the RegEx would have to find |archiveurl=http://[web|wayback].archive.org/)

Event Timeline

bender235 raised the priority of this task from to Medium.
bender235 updated the task description. (Show Details)
bender235 added a project: AutoWikiBrowser.
bender235 set Security to None.

English-language-centrism. enWP is not the font of all knowledge and the determiner of all linking styles. As such that this an interfering fix that is unnecessary and meets personal preference, not a necessary outcome.

@Billinghurst - Are you saying that this fix would not be appropriate on wikis outside of Wikipedia (any language)? Or would not be appropriate on Wikipedias except for enWP? Or are you also saying that this would not be appropriate for enWP? If the latter, could you please expand on how it would be "interfering"? Thanks!

Please provide a link to a centralised discussion confirming that this change has consensus for wikipedia (en-wp at least). VPT may establish technical merit, I would like to see WP:CITE etc. discussions being clear HTTPS links are preferred in these cases.

There are four sites mentioned but not scope/examples for all four. It says cite templates only yet there's no explanation why this would not apply to other external links, links generated by infoboxes etc.

If were to implement such a change we need more clarity.

@Billinghurst What makes you think this kind of fix would interfere with an "personal [language] preference"? Language is not affected at all by changing the link protocol from HTTP to HTTPS.

@Rjwilmsi Back in 2014, when Wikipedia still gave the option between HTTP or HTTPS access, we had a discussion that concluded we should make people exit Wikipedia on the same protocol they entered (i.e., protocol-relative links), provided that a given website offers both HTTP and HTTPS. Now that Wikimedia projects are all HTTPS-by-default, this implies we should have HTTPS links where ever possible.

@bender235. You may wish to reread what I wrote, "personal preference" != special:preferences.

Firstly, I hardly find that simple conversation at WP:VPT can be called a consensus. What we do by encouragement, and what we do by force, and by our opinions are different things. Before you go and enforce into AWB a means to change links, there should be amendments to the editing guidelines through the MoS that the links should be styled as https in preference to http if that option exists. Such a change should be through a Request for Comment process. [Sure that makes it harder, but it makes it open and it makes it through a proper consensus. Don't shortcut the people.]

Secondly, rewriting urls from the form that they are added to a new/altered domain form, as you have slipped into the phab ticket, was not part of the discussion, and should not be undertaken. Your personal preference for a domain is not relevant, there is no requirement for any of these links to be in .com, away from the originally added domain variation. It is an unnecessary change, a change that those who entered the link could rightly take umbrage, and technically introducing the ability for error.

The AWB product is used across all the wikipedias, and not just English Wikipedia,. In fact it is used across all the sister wikis. Expressed requirements for change should take into consideration of all the wikis, not just your suggestion at enWP:WP:WPT. While AWB is used as "user beware, user be responsible for edits" changes that have the potential to break things should be avoided. Here are suggesting adding a change for a genfix that is not actually fix at all, just a change based on a preference. The change if used outside of enWP could break other wikis rules, and as such I believe it is a bad idea for genfix. Have it as a plug-in *if* you get it through an RFC, but it is not a fix, and it is definitely not a genfix.

[Noting all this is not about whether use of https or http is preferred, it is about the process for getting a change, and whether the change should happen. This is about going through the right processes for community consensus. not sneaking in a policy change through the backdoor.]

For what it is worth, templates are different, and would usually present a specific configuration which people can choose to use, or not, so I am not certain of the pertinency of the comment. That are what they are.

@Billinghurst As for the "consensus", no one was referring to the ongoing discussion on WP:VPT. I was instead pointing at the year-old discussion at WP:VPP.

Second, switching to "books.google.com" from "books.google.[de|it|es|ca|pt|...]" is not about personal preference. Before complaining about this, please take time to understand the issue. If you browse to "books.google.com" from anywhere outside the US, you will be redirected to your country's language version by Google. But if you browse says "books.google.it" outside of Italy, this won't happen. So when ever a non-US-resident adds a Google Books link on Wikipedia, it'll most likely be with the top-level domain of his/her country version. But just because an Italian person added the link doesn't mean only Italians will click the link on Wikipedia. Hence, the link should be fixed to the "generic" version before the language redirect.

But if this issue is this confusing and unacceptable to you, I am willing to drop it. It was only secondary to the issue at question here, anyways.

  1. WP:VPP is a closed discussion without consensus. For something to appear in genfix, I would propose that it would have to comply with direction from [[w:en:WP:MoS]] until that point it is personal preference. [The setting of https for the foundation for its use, is not about enforcing its policy elsewhere. I am trying to avoid having that policy decision in a side phabricator ticket, and should not be used a reason to get a genfix, as said genfix should be about complying with MoS only, the lowest common factor, not the highest level impost, not political issues]

I don't live in the US, and I get perfectly formed urls at the link as by articles, no redirects at all. Retested today for .it .es .de links via special:linksearch. I do not see a consensus for this general fix.

@Billinghurst: The WP:VPP closing statement says there was a "clear consensus." How much more does it need for you to recognize?

Josve05a changed the task status from Open to Stalled.Sep 8 2015, 5:11 PM

Marking as stalled until the two of you have figured this one out. The developers of AWB will not be implementing this as long as there is disruption discussion regarding if it should even be part of AWB, therefore by definition, it is stalled.

A limited discussion with that summation at a single wiki, or a single
Wikipedia, does not and should not set the linking and updating of
genfixes.

AWB does genfixes for how many wikipedias? Have you asked each of them?
Have you asked any others? How about just some of the other big wikis?

Show me where the discussion has been broader and asked more than the
watchers of that page? Show me where it has been incorporated into policy.
Show me where it has made it into the MoS. It hasn't. Until that has
occurred, all you have is an indication of a small group at a page at one
wikipedia, nothing more, and from 18 months ago. That is not a consensus to
change every Google and archive.org link.

The current discussion is about policy/consensus for the change, not the technicalities of the implementation. So I think the policy discussion should be on-wiki/on meta, not on a ticket, and we should return here if/when clear consensus is reached for change.

Why would we not simply make this apply to EN only? We can always change it later.

For webarchive, because have https, i request to add changing from http to https. I resolved that using python on Serbian Wikipedia. Https is security protocol. I requested also to add https when weblinkchecker script report dead link on talk page (https://web.archive.org), no http://web.archive.org

Aklapper changed the task status from Stalled to Open.May 21 2020, 10:43 AM

Just to let you guys no, I opened an RfC on this issue.

https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(proposals)/Archive_127#RfC:_Should_we_convert_existing_Google_and_Internet_Archive_links_to_HTTPS%3F says Closing comment: There is near-unanimous consensus in favor of converting the mentioned links to a secure HTTPS connection.

I have no idea what that means for this ticket, but ticket should not be in "stalled" status for five years, hence reopening. :) (Feel free to decline, etc.)

Aklapper lowered the priority of this task from Medium to Low.Dec 14 2022, 1:13 PM
Aklapper changed the subtype of this task from "Task" to "Feature Request".
Aklapper moved this task from User input needed to Backlog on the AutoWikiBrowser board.