Ask the Mods Important Site Questions (V2)

Just wondering something:

I noticed that some user’s profile pictures show a name or date when you view it in a new tab, such as lwut’s profile picture. Just wondering if there’s a way you can modify the name when choosing a new profile picture or if that was an old effect that is no longer in process?

If it were possible to include the file name I would like to try it 😀

Explain? _(please)

@NightStalker
By which they mean, “explain why that post was evaluated for Rule #3 and it was decided it did not violate it”

Yeah for real, that’s bullshit– not only is it a clear Obligatory Pony violation, the uploader straight up admitted it.

When you go to the forums, nothing is showing up under the “last post” section for the “Uploader Discussion” forum despite the last post their being 4 days ago.

Is this a bug or is nothing showing in the “last post” section for that forum supposed to be intentional?

@NightStalker
I wish I could explain. Right now staff agrees it’s self-evidently in violation of #3 and it’s been re-deleted. The mod note on the image merely said it had been restored “as per second look”. That wording is found on other images as well, suggesting this image may have been accidentally targeted by a batch action, but the site’s UI doesn’t support looking up mod actions that far in the past. We’re looking into it tho, and should be a bit wiser once the acolytes have erected big enough bonfires for the blood sacrifice ceremony back-end search.

@icicle wicicle 1517
Intentional, when the most recent post was deleted spam.

@saby
Words cannot express my relief. Thank you.

So the description for the “tv meme” tag says:
Note:using this template does not make something pony-related, and non-pony stuff will be deleted.
Yet there’s tons of TV meme images with non-pony stuff still up. Why is that?

@Blissful690
Because they’re old.

Nothing deeper than that.

@Blissful690
Old exploitable memes are preserved for their historical value. If you find new ones, please report them so staff can act accordingly.

The image upload form says

Can we please change this, because apparently it’s too complex for some people to understand. Explain in simpler words that “3+” does not mean three and only three.

@JP
I’d wager the problem is overwhelmingly in attitude, not comprehension.

@saby
Unfortunately, you’re right.

:-(

Include a feature that generates a low priority auto-report if an image is uploaded with five or fewer tags, so a mod can dope-slap repeat offenders.

@TexasUberAlles
If there’s a way to do a query for images where tag_qty <= 4, then one wouldn’t have to deal with multiple reports, just run that on a slow day.

@𝗪𝖎𝖗𝖊𝖇𝖆𝖑𝖑
There is a way, use the tag_count field. It’s listed on the supported fields page.

Here’s a working search for images with only three tags.

And some images have more than 500 tags.

@TexasUberAlles
This makes my anatomy tingle. Something similar has been suggested in a recent staff thread, and this would compliment those suggestions nicely.

@𝗪𝖎𝖗𝖊𝖇𝖆𝖑𝖑
That only works the fresh, hot second when some helpful user hasn’t added a few extra tags. Automation should be next to trivial in this case, and also magnitudes more efficient.

@JP
Oh that’s quite handy. I was aware of tricks such as prepending minus to filter out certain keywords, and score.gt:5 (greater than 5), but there are several like tag_count (exact) or tag_count.lte (less than or equal to) that I hadn’t investigated. Thanks!

@saby
Those fiendish helpful users! I see what you mean.
It also sounds like your report handling system is fairly fast to work with.
_{(I’ve been somewhat scarred by one of our older ticketing systems at work - 60+ seconds just to open and assign a ticket, and it fights you if you try to open tickets in different browser tabs.)}

Why pending approval from a staff member after uploading an image by anonymous can sometimes last a few hours?

@Background Pony #0414
Well, just talking about your last upload, it’s expressed as a video in the system, and two times I checked the queue it was still being processed.

We can’t approve stuff until it’s finished processing (or otherwise can be seen in its entirety) so I didn’t approve it.

Later, I checked back, and it was finished so I approved it.

Plus, no one is sitting at a desk watching the approval queue. We all just swing by it whenever we have a chance to see what’s there.

So, for me, I saw it three times over maybe an hour and it wasn’t in an approvable state until my last time through.

Does that help explain how it works and why sometimes it might take awhile?

Also … it seems you might have a profile. If you do, just upload the image logged on. You can even upload it anonymously. Then it skips the approval queue.

@Ciaran
>>3140014 is still waiting for approval for 6 hours already, it was never that long.

Some characters are rendered in Derpibooru tag URLs are HTML-encoded sequences (in which the % character is further escaped as %25), but I can’t see much of a pattern over which ones get encoded and which ones remain unencoded. This seems particularly confusing regarding Chinese characters, it seems that one Unicode block gets encoded whereas the next one doesn’t, which screws things up for my attempts at writing userscripts.

Two contrasting examples for the sake of illustration:

Not to mention the board-specific old escaping like : becoming -colon-, tho I’m used to these.

Is there an official list of which of these 3 things is done to each character?

Not a mod, but can help:

@OneMoreAnonymous
The list of character-to-string substitutions is in slug.ex. They are then encoded with Elixir’s URI encode function which I do not know the specifics about how it works, but one time when I tried to replicate Derpibooru’s tag encoding, I ended up with encodeURIComponent(str).replace(/[!'()*]/g, c => '%' + c.charCodeAt(0).toString(16)) in JavaScript (taken from Stack Overflow)

Note that browsers may perform a silent conversion between the encoded form and the human-readable display of characters: for example, for me in Chrome, accessing your second link displays the Chinese characters in the URL bar, but the network tab of the developer tools reveals that a request was made to /tags/artist-colon-%E7%94%B0%E5%9B%AD%E9%94%84%E4%B8%B2%E5%AD%90

However, there is a wrinkle which I believe makes it impossible to programmatically generate a tag’s URL: tags first used before this patch in 2020 were double-encoded, and won’t work if single-encoded, with no way to know which way a tag is without either just looking it up through the API (eg. artist:未知的生物, artist:田园锄串子) or to cache a list of all affected legacy tags, which I have done but is 56 times too many characters to post in the forums.

Is it possible to search multiple galleries? For example, I’ve tried searching for gallery_id:15111,gallery_id:21377 but no results come back.