14

One of the steps on Stack Exchange's road to HTTPS is the replacing of embedded http:// images with secure ones. This has been done automatically for a lot of images, but there are some that can't be automatically fixed. These were rewritten in the HTML (but not the Markdown source) to be text links instead of image embeds. It would be good to go through the affected posts and make images display correctly, or remove the link debris if they can't be restored.

Since there seems to be a problem with the CrowdCrafting task's data, I used this SEDE query (slightly adjusted from this version by Jefromi) to find a few hundred posts that may require our attention.

To continually track our progress, we have a Google spreadsheet. (Even though all the links have /q/, answers are included too.) The Status column is editable by anyone, so please note your work there as you go along.

As usual when doing big batches of edits, try to avoid flooding the front page, and please try to fix all obvious problems before submitting an edit.

  • 4
    Could this not be done at the database level? BTW, I'm playing devil's advocate here. – Burgi Mar 12 '17 at 21:07
  • 1
    @Burgi Unfortunately, the remaining images are not accessible over HTTPS. Apparently, Stack Exchange can't automatically upload HTTP images to an HTTPS source for licensing reasons. For more discussion of the image transition, see this MSE post and its comments. (I do agree that it would be great if this could be done automatically for us!) – Ben N Mar 12 '17 at 21:26
  • I've just discovered that not all the http:// in that list directly relate to images. Perhaps a tweak to the query is needed to see if the URL ends in jpg, png or gif? – Burgi Mar 12 '17 at 22:15
  • 1
    @Burgi Ah, if ![ appears in the post, the query assumes it's an image, even if it's actually escaped. My SQL skills are not strong enough to deal with that, but I would be happy to link any improved query! – Ben N Mar 12 '17 at 22:21
  • Feels like a good time to do a free for all tag cleanup. Let me consult the other mods and see if there's any violent objections and I'll post a meta post for that. – Journeyman Geek Mar 12 '17 at 23:16
  • 1
    what is the guidance for 404s? Nuke or leave in place? – Burgi Mar 13 '17 at 0:24
  • 4
    If you can find an appropriate/logical alternative, go ahead and include it. If you can edit out the image link and the post makes sense, go ahead. If its trivial to generate a new image, go ahead. Otherwise, I guess we need to go by a case by case basis – Journeyman Geek Mar 13 '17 at 0:53
  • @JourneymanGeek Is this something a low rep user can assist with? Or, would that just make more work by loading the review queue with my edits? – Gypsy Spellweaver Mar 13 '17 at 1:06
  • 1
    @GypsySpellweaver An edit approval is faster, then doing the edit, feel free to help – Ramhound Mar 13 '17 at 1:27
  • 2
    Yes please! Do link back here in the edit reasons so people know what's happening and do other appropriate fixes as needed – Journeyman Geek Mar 13 '17 at 1:32
  • Might want to add a filter to the spreadsheet. I tried to do so but visible only to me – Sathyajith Bhat Mar 13 '17 at 4:44
  • Suggest working from the spreadsheet. The query has a lot of false positives, including posts that have been fixed. People will spend a lot of time revisiting posts that have already been reviewed. Updating the status column on the spreadsheet will save people a lot of time. – fixer1234 Mar 14 '17 at 10:09
  • 1
    @Sathya Good idea, done. The protection should still be working as before (with the Status column in its new place). – Ben N Mar 15 '17 at 18:48
  • 1
    @Hashim The SE-hosted image links have all been automatically fixed for us. What's left are those that aren't on SE and aren't accessible over HTTPS - we can reupload the existing HTTP-only ones to Imgur. There are also plenty that are broken over both protocols; these need to be tracked down if possible, or replaced. – Ben N Mar 20 '17 at 2:24
  • 1
    @Hashim I think the HTTP → HTTPS URL changes for Imgur are automatically handled, so there shouldn't be any such instances in the spreadsheet. Once you fix a post, please update the Status column to say "Fixed"; there are a few other statuses that people have been using for other cases. – Ben N Mar 20 '17 at 22:32
8

Summing up the comments

Here's what we essentially need.

High reputation users should probably keep an eye on the edit queue -and improve and accept rather than reject edits related to this fix up. If you're a lower rep user, feel free to edit, and link back here (and if you see anything that needs to be fixed up along with images, please do so). This feels like a great opportunity to get people to fix things.

If an image can't be found, and a reasonable best effort with the waybackmachine doesn't help, feel free to replace the image with a suitable replacement if it's 'obvious', or remove it if the post can manage without it. If neither helps, leave it alone first (for now!) and we can work out what to do later.

I'd add its an excellent chance for people to work towards copy editor and steward badges.

I'm going to see if the other mods, and the community are on board with having an (extended?) period for cleanups like the tag cleanup weekend we had a little back. We should probably take the chance to handle the tag cleanup backlogs as well. In theory, don't flood the front page for now, but I would be slightly more inclined to be sympathetic if it happens than I would normally be.

Non https stack.imgur links should be autofixed - but there's probably no hard in swapping the protocol there

  • Haven't been here long enough to know what "flood the front page" means, nor how to not do it. At a guess, an edited post become "current" again, and moves to the front of the list? If so, then it seems to make sense to work from oldest-to-newest in the fix. That way, as the newer posts get edited and move up, they naturally move the older ones that moved up because of the edit back down to something that more resembles their prior position. I think? – Gypsy Spellweaver Mar 13 '17 at 4:20
  • Yeah, and a lot of people only use the main page, so bumping kinda disrupts the way we find questions to answer. Typically the advice is not to do too many edits at once - though we work around that by declaring open season on necessary edits periodically. – Journeyman Geek Mar 13 '17 at 4:54
  • I'm slow enough that my work won't disrupt very much at all :) Too bad there isn't a tool for mods to override that behavior. Not that the developers don't have enough to do already. – Gypsy Spellweaver Mar 13 '17 at 4:58
  • eh. We can solve this as a social rather than a technical problem. That's why we have mods after all, handling exceptions. – Journeyman Geek Mar 13 '17 at 5:02
  • Ok, maybe this is better to ask in chat, only I don't know how/where, so... Reviewing one of the posts to edit I found five dead links in the text. Removing the links won't change the text, or answer, at all. The static versions of the links are in Wayback, but they server no purpose because they don't link further into material that the post refers to. Do I remove the dead links, and keep the text, while fixing the image link? – Gypsy Spellweaver Mar 13 '17 at 5:57
  • 1
    I'd go for removing the links if non essential, or sticking a way back machine link if it helps. For now, I suppose root access is the place to go. I practically just started working on getting this started a few hours ago. - so I'm still dekinking all the kinks. – Journeyman Geek Mar 13 '17 at 6:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .