Feature suggestions and discussion [READ THE FIRST POST]

@OldTalesChangeStyle  
The intent is commendable and the request is legitimate.  
I hope the developers will take it under consideration, even though you have to admit that you’d basically use Derpibooru’s CPU power to make your bot work (in other words: develop it so it won’t abuse the API).

@BigMax
 
Right, and I’m well aware of that (thus the “along with reasonable rate limiters if needed” qualifier in the previous post). Reddit has its own API guidelines that determine how many requests can be made per a certain amount of time, and I imagine that Derpibooru could construct a similiar policy, or alternatively control the API keys they give out for the reverse image search, perhaps by having developers apply for them.

Goddamnit we need some sort of spellcheck option or correction. I can’t tell how often I haven’t been able to find an artist cause I can’t spell his name. Having to search through the booru until you find a picture of the artist so you can get into the tag gets really tedious.

@Background Pony #5587  
We should have live suggestions, like when you add tags. And it should fuzzy match, so “chse” would still match “chest fluff”. (There’s this, but IDK if it’s better to call it from Ruby or to reimplement it.)

@Background Pony #BC7C  
We have Elasticsearch backing most of our pages so can call out to do suggestions through that pretty readily. It’s just not done/exposed anywhere. We probably could without too much fuss, though it’d involve breaking complex queries up and doing quite a few suggestions. I think I’d be right in saying that doing it for simple cases only or just supporting it on tag searches would be a major improvement, no?

“breaking complex queries up and doing quite a few suggestions”  
IDK if I understand. It could work like this: user starts to type a tag -> offer live fuzzy matched completions for this tag. User types stuff like brackets or operators -> ignore it.
 
But even if you did this for simple searches and tag completion, it would be great.

The first post said you guys wouldn’t add an option to hide people’s posts because it taxes the server, but also menctions scripts one can use to filter out annoying people.
 
May I ask some help with that? I’m tired of people who only visits this place and the board to insult and antagonize others, so a personal filter would do wonders here.

@Anonycat  
iirc someone made a greasemonkey script that hides divs based on the profile ID of the commenter.
 
That said, if you see someone whose only activity is to be a little shitemeister, report them.

@Anonycat  
I think that’s B.O.R.E.D. you want.

@Solitude  
Danke.  
I was really tired of the “Stop whinning, stop thinking, stop being a hater, your opinion is worth nothing, shut up, shut up…” kind of toxic posters.

@Background Pony #BC7C  
Apache Lucene allows for fuzzy string matching with a Damerau–Levenshtein distance of up to 20% of the queried string length, so it’s actually quite possible to do.

Out of curiosity, would it be possible feasible to embed all of the tags on an image as keyword metadata, so that they’re all retained when an image is downloaded?

@Silvercloud  
Enough of them are already included in the images’ filenames to make them archive-ready.

@Silvercloud  
This has been asked before. Some formats (JPEG, PNG and SVG) have support for arbitrary metadata, but unfortunately most programs (even dedicated image viewers and editors) are unable to show or use that data in any way. With JPEGs it might work fine, as EXIF keywords are widely supported, but for PNG and SVG, I have my doubts. GIF supports comments, but I don’t know if they’re suitable for this kind of (ab)use.
 
So yes, it can be done, but it wouldn’t work very well.

@JP
 
This is basically it. We could write out XMPs or other metadata files, but very few tools can actually work with them and they tend to be at the pro end of the spectrum (Adobe, Avid products etc). We provide image metadata via the API (append .json to any image ID, eg /images/1234.json) and if you want to embed that data it’s a fairly simple operation to ping that endpoint, grab the data, write the metadata you want in the format you want. We don’t do it ourselves because there’s no one right answer for how that’s done for all use-cases; we just provide the data in a sensible format to let others do what best suits their requirements.

@Clover the Clever  
Never let it be said that I’m not willing to put in some effort to get what I’m after… after entirely too many hours of work, I now have a Finder folder action that will ping the API for tag data whenever an image from derpicdn.net hits my downloads folder, add the tags to the file’s metadata (using OpenMeta), and then move it to my image library.
 
In other words, I am a total nerd. Thanks for the pointer on the API :)

@Silvercloud  
I don’t use a Mac anyway, but you could post source for the future readers.

The site should automatically reject an attempt to upload an image by URL if that URL contains “derpicdn.net”. For example >>832838 (merged)

@VladimirMacHolzraum  
It would be better to just create a blacklist of domains from which you cannot upload from.

Uploaders who post Season 5 spoilers without the spoiler:s05 tag should be banned until Season 5 starts.

Some tweaks and cleanup under the image in the image page.
 

We need a way to report anonymous uploaders, because reporting the pic doesn’t fucking work.

@dt  
Can you elaborate on “doesn’t fucking work” in the bug report thread so we can fix that?
 
Or are you just complaining about mods again?

@Clover the Clever  
He’s complaining about mods again.

@boogstersu2  
While iconizing more stuff might be nice to save space, removing borders would be a no, as would any sort of round edges.
 
Also keep in mind that you’re looking at the site through the eyes of someone that’s been here for years. You have to keep in mind new users landing on the site for the first time.