[ / / / / / / / / / / / / / ] [ dir / random / cloveros / cuteboys / erp / fast / hydrus / in / strek / tftg ]

/hydrus/ - Hydrus Network

Bug reports, feature requests, and other discussion for the hydrus network.
Name
Email
Subject
Comment *
File
Password (Randomized for file and post deletion; you may also set your own.)
Archive
* = required field[▶ Show post options & limits]
Confused? See the FAQ.
Embed
(replaces files and can be used instead)
Voice recorder Show voice recorder

(the Stop button will be clickable 5 seconds after you press Record)
Options

Allowed file types:jpg, jpeg, gif, png, webm, mp4, swf, pdf
Max filesize is 16 MB.
Max image dimensions are 15000 x 15000.
You may upload 5 per post.

New user? Start here ---> http://hydrusnetwork.github.io/hydrus/

Experienced user with a bit of cash who wants to help out? ---> Patreon

Current to-do list has: 2,017 items

Current big job: Catching up on Qt, MPV, tag work, and small jobs. New poll once things have calmed down.

HookTube embed. Click on thumbnail to play.

3d1d90  No.8979

windows

zip: https://github.com/hydrusnetwork/hydrus/releases/download/v309/Hydrus.Network.309.-.Windows.-.Extract.only.zip

exe: https://github.com/hydrusnetwork/hydrus/releases/download/v309/Hydrus.Network.309.-.Windows.-.Installer.exe

os x

app: https://github.com/hydrusnetwork/hydrus/releases/download/v309/Hydrus.Network.309.-.OS.X.-.App.dmg

tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v309/Hydrus.Network.309.-.OS.X.-.Extract.only.tar.gz

linux

tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v309/Hydrus.Network.309.-.Linux.-.Executable.tar.gz

source

tar.gz: https://github.com/hydrusnetwork/hydrus/archive/v309.tar.gz

I had a good week with a bunch of unusual work.

tumblr GDPR

When GDPR hit, tumblr introduced a click-through page for all European Union and European Economic Area users. Unfortunately, this page applies (likely unintentionally) to their old JSON API as well as regular pages, so it broke the current hydrus tumblr downloader.

I have written a simple 'login script' to perform this click-through manually, so if you are an EU/EEA user, please hit network->DEBUG: misc->tumblr click-through. The magic cookie you get lasts a year, so we have plenty of time to come up with a better solution. I understand this click-through event is linked to tumblr logins, so as I roll out a proper login system, we'll have more control here.

I live in the US, so testing for this was not super convenient. It appeared to work for a couple of EU/EEA users, but it may not for certain countries or may suddenly no longer be needed if tumblr changes how or where it applies to their CDN. I don't know enough about the implementation or the law's details to know for sure, so please send your feedback. This–and any other sites that might be forced to switch in the coming weeks–is likely something we'll just have to keep a continuing eye on.

video thumbnails

Videos (but not gif yet) now generate their thumbnails 35% in by default! You can also change the 35% value under options->media. You should see far fewer black-frame thumbnails for your videos now!

This was a highly requested feature. While I had originally planned to wait for a full overhaul of the system that would introduce animated thumbnails and maybe even an 'interestingness' scanner to pick the 'best' frame and discard blank ones entirely, after thinking about it more and talking with several users the past few months, I decided to have a brief look at the actual code and saw I could hack in some simple support. I cannot promise it works for files with unusual/inaccurate frame counts, so please link/send in any video files that error out.

I will eventually write an automated maintenance routine to regenerate all your old videos' thumbnails (and some other stuff that still needs a rescan) in idle time, but if you can't wait, turn on help->advanced mode and right click the selection of files you want to regenerate–there should be an option at the bottom to reparse and rethumbnail them.

downloader stuff

I've added parsers for inkbunny and gelbooru 0.2.0 (by default in the client, that means rule34.xxx, tbib, and xbooru), so they now support drag-and-drop and, for gelbooru, parse a little more info than before.

The multiple watcher now has a pause button, better sorting, and will prefer to accept url drag-and-drops on the current multiple watcher page, if multiple are open but one is currently selected.

The way the client performs its 'Have I seen this URL before?' test is now much more strict. False positives should now be far less common when some booru attaches a bare username/artist twitter/patreon/whatever URL as the source for a file. If you have been hit by this, let me know if it all magically fixes this week if you 'try again' the affected URLs, and if it doesn't, please let me know the details. I am still working on this and will roll out more tools to control url-checking behaviour.

Advanced users: The HTML parsing formula's tag rules can now search for tags that have certain 'string' content, like 'Original image'. This is very helpful in certain situations–check out my new gelbooru 0.2.0 file url content parser if you want to see an example!

siblings and parents logic improve

Much like the video thumbnails above, I decided to hack-in some better sibling and parents logic this week rather than wait for a neater complete overhaul.

So: now, if you create a new parent pair c->p, c will now also get any parents of p, and and parents of them (i.e. grandparents are now added to children recursively). Also, if you add a sibling pair a->b, a and all of a's siblings will get all the parents and grandparents (etc…) of b and b's siblings, and vice versa. Basically, parents now add recursively and all the siblings get all the parents they are supposed to. It isn't perfect yet, but it is better.

And again, I expect to write a maintenance routine in the future to retroactively fill in the gaps here.

Linux stability

I have put a bunch of time into improving Linux stability, which has, on-and-off, been randomly horrible since the wx update. I haven't caught everything, but I think things are better. Please continue to give feedback–whether you notice any difference at all, and if any activity in particular is still likely to cause a crash a few minutes later.

full list

- wrote a fix for the tumblr GDPR issue under _network->DEBUG: misc->do tumblr GDPR click-through_. you will also get a popup about this on update

- the tumblr downloader will try to detect the GDPR problem and present a similar popup guiding you to the GDPR click-through solution

- the client and server now generate video (but not gif yet) thumbnails 35% in by default. the client can now change this percentage value under options->media. this was highly requested and was being put off for a longer rewrite, but I figured out a simple way to hack it in. please let me know if you get failures

- on adding a parent, all files with the child tag will now also get all applicable grandparents (with no limit on recursive generations and dealing with accidental loops)

- on adding a sibling, all files with any of the siblings will now also get all applicable parents and grandparents for the whole group. a maintenance call to retroactively fill in the sibling/parent gaps that are now filled will also come soon

- this logic still does not apply in cross-service situations, which _will_ likely have to wait for a big data/gui overhaul and us figuring out what we actually want here

- added a simple pause/play button to the multiple watcher

- if the multiple watcher is set to catch watchable url drag and drop events and the current page is a multiple watcher, this current page will catch those new urls (as opposed to the _leftmost_ multiple watcher)

- improved some thread unpause logic which was failing to lock pause during 404 status

- the multiple watcher should now ignore case when it sorts by subject

- added url class and file page parser for inkbunny (so this site is now supported in drag and drop!). it fetches creator tag, some artist-made unnamespaced tags, source time, and md5

- added file page parser for gelbooru 0.2.0, which by default works for rule34.xxx, tbib, xbooru but certainly should work for a bunch of others. it fetches source time and source url

- html formula parsing rules can now additionally test the tag 'string' using a standard StringMatch object. this greatly helps to parse otherwise indistinguishable 'a' tags that have string 'Original image' and so on

- the 'have I seen this url's file before?' pre-import test is now much more strict and will cause fewer accidental false-positive 'already in db'/'deleted' results:

- the url pre-import test now does not trust source urls if they do not have a url class

- the url pre-import test now no longer trusts urls that are supposed to only be mapped to one file but are actually mapped to multiple

- this url pre-import test now treats url-classless original post urls and intended file urls with a special level of trust

- urls are now stored in the db in a more powerful and in-future easily searchable way–your db will take a moment to convert to the new format on update

- did some prep work for multi-file post urls (like pixiv manga) but did not have time to finish it

- the filename tagging options panel (in the 'add tags based on filename' of file import dialog and import folder dialog) now updates its tags/list 0.5s after the last change event, which means typing on a giant list will not cause megalag

- improved stability of some client-screen coordinate conversion

- misc bmp handling stability improvements

- improved some parsing ui stability when example data gets set after the dialog is closed

- improved some misc dialog close stability

- converted all but one final ui update timer to the new job scheduling system

- there are still problems with linux stability–I will continue to work on it

- an ugly (but basically harmless) shutdown exception sometimes caused by Animations being a bit slow on deleting their underlying bmps _should_ be fixed

- the export files dialog now generates its paths in sort order, meaning (1), (2) de-dupe filename suffixes should now be generated nicely in order

- the network domain manager should now always chase API URL links to get the right parser

- made some 'the db is broke, let's try to fix it' tag recovery code more forgiving

- misc improvements to some media indexing backend, which may fix some unusual session ghost files

- fixed the 'sure it is ok to close this importing page' dialog to also veto on a 'cancel' event, rather than just a 'no'

- added a guide to database_migration.html on how to move the db from just an HDD to straddle both an SSD and HDD.

- cleaned up the help->debug menu a bunch

- added run fast/slow memory maintenance calls to help->debug->data actions

- misc cleanup

next week

I would like to take the week from the 6th to 13th easy so I can shitpost E3, so I'd like to concentrate on tying up loose ends. I'd love to get pixiv manga pages working along with other multi-page Post URLs like tweets and Artstation posts–I did a little on this this week, but then ten other things like GDPR swamped me.

____________________________
Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

37bd07  No.8981

>>8979

haven't installed it yet, but an idea on the video thumbnails, would it be possible to make an arbitrary thumbnail point? as in the user defines a frame, and that is the thumbnail?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

47b935  No.8982

>>8979

Thanks Dev!

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

13a5ae  No.8983

>>8979

nice.

I have a minor request: Could you make it so the first file presented while in a dupe filter is configurable? Right now it seems random, I think it would be more useful to sort by filesize or resolution, etc.

For example: say I have two images that visually are exactly the same, but one has a larger filesize. When I go into the duplicates filter, there's no way to know if the smaller one would appear first or if I would have to click next.

Personally I'd like to see the smaller one first, but the option for either would be nice.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

35417e  No.8986

>>8979

Minor Request: Software Module Unit Testing?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

735c50  No.8988

File: 16410c760546809⋯.gif (448.05 KB, 409x226, 409:226, 16410c760546809f685d1489b6….gif)

>I would like to take the week from the 6th to 13th easy so I can shitpost E3

Based dev.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4dcf08  No.8992

>>8979

great stuff dev! Thanks again for everything! Did we get the thumbnail generation for OS X server working in this update?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

d7303f  No.8993

File: 2db2f92b0bef109⋯.jpg (60.57 KB, 832x690, 416:345, breath.jpg)

>>8979

>I would like to take the week from the 6th to 13th easy so I can shitpost E3

we can do with no new release for a week hydev, take a vacation man, you've given us a lot already.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

bcc966  No.8997

File: 46875883e0bf48d⋯.png (17.75 KB, 523x364, 523:364, 46875883e0bf48d4ef4711ba6b….png)

i have a problem. i haven't updated my hydrus in a long time and then i decided to. i thought i remembered that all you had to do was rewrite every file with the new version. i did that and then it wouldn't start. it gave me an error saying missing master boot files and such, with the last box appearing as pic related.

i know the db folder stores all the images. so i extracted the new build again and started it, worked fine. then i dragged my db folder over, and it's that same error. i imported all my pictures but.. i have no tags anymore. i want my tags back.

whats the problem, what can i do? what is the proper way to update for next time?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

47b935  No.8998

>>8997

>what is the proper way to update for next time?

doing it week to week you dingus. 0 problems ever.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

bcc966  No.8999

>>8998

thats not how it should work. you are essentially telling me to download this shit every wednesday when he updates or stick with an oder version forever. and i don't always have the time for that.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

47b935  No.9000

>>8999

That is how it works when some weeks have major changeups that the db needs to be updated for and you missed the entire update window. I update every 1-3 weeks but I always run each of the updates in order.

All I know is every time there's somebody trying to jump 20, 50 or 100 versions between updates their db winds up broken. At the very least they could run 10's just so they'd have a better idea where the screw-up was.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

bcc966  No.9001

>>9000

so all i have to do is copy and paste the new build over to the last one right, like i tried to do before? is that how people still do it?

and what about all my tags? are they gone?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

66dc59  No.9004

>>8997

>what is the proper way to update for next time?

Go to the Releases page and upgrade 1 version at a time starting from your version until you reach the most recent version. This way incremental database updates, since a lot is still a WIP, doesn't corrupt your database. Don't try jumping multiple versions and expect it to work like other software that doesn't maintain (or need to restructure) an internal database.

https://github.com/hydrusnetwork/hydrus/releases/

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

34afb4  No.9006

>>8979

can you explain a little more indepth how the inkbunny downloads work now?

im copying links into the "url downloader" just like the example in the parser specifies and all i get is "couldn't parse any data", or it gives me a bunch of thumbnails if i do it through the simple downloader

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

b9d923  No.9015

>>8997

Don't listen to these chumps, just backup your database and shit, delete all of the files in the Hydrus directory other than the "db" directory, and paste the files over again, if you don't see any obvious fuckups, then you should be alright

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

bcc966  No.9016

>>9015

nope. its fucked. but i just realized i had a backup so im good. ill just update one at a time i guess

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

b3c3da  No.9018

I did a mime->webm search to regenerate all the thumbnails at once, but one webm gets the error message "Unable to render that video! Please send it to hydrus dev so he can look at it!" which just aborts the entire process. And it doesn't even tell me which file. So that's useless.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

34afb4  No.9021

File: d956de96ff7f6bc⋯.jpg (25.12 KB, 255x179, 255:179, Guilmon.jpg)

i wanted to do a little report here on my experience with the IB parser, i dont think its working as intended.

first i had some trouble with the PHPSESSID, it was different in hydrus than it was on my browsers cookies, so i manually corrected that because you need an account (just like pixiv) to see anything that isn't rated safe, otherwise you get a "could not parse any data!" error, logically so, as inkbunny doesn't allow an outsider to see NSFW content. I did so by copying the content of my browsers inkbunny PHPSESSID into the hydrus IB-related PHPSESSID, overwriting the one that was already there.

so, with that fixed, i tried to use a link in the URL downloader just like the provided examples in the parser, but this time i looked for a multi-image submission.

IB just like Pixiv allows people to make a single submission with several pieces of media on it, can be multiple gifs, flashes, webms, pictures, or text. The result was that only the first picture of the 'set' is downloaded.

there's always the possibility that i'm being retarded and just using it wrong, but i'd like your opinion on that

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

3d1d90  No.9024

File: b5ba70a9db6e384⋯.png (339.34 KB, 656x435, 656:435, b5ba70a9db6e38498b598c3349….png)

>>8981

I'll do this at some point, also to help associate nicer thumbs with pdfs, swfs, and cbrs and so on, but it'll need the cleverer thumbnail system I've been planning.

>>8983

If I remember right, it shows a rough idea of which file is 'best' first. The different things it lists at the top, like 'this has higher resolution' each have a scrappy weight I threw together, and the two files have their 'this is good weight' summed and compared when they are first loaded, with the higher coming first.

Exposing these weights to the user is doable (e.g. you could set them all to 0 and then higher res to -1 so the smaller res always presents first), so I will make a job for this. I'm overloaded with downloader stuff, so I can't get to it fast.

>>8986

Improving unit testing is an ongoing job, but one I find difficult just because of my workflow. There are a variety of thrown-together unit tests that I run every week, but the actual code coverage is tiny. I am concentrating on downloader stuff, which has sprawled into a much larger job than I expected, but when that eases off, I'd like to put a bit more time into overall code cleaning and unit test improvements.

>>8992

Not yet–I'm afraid my schedule fell apart this past week. This one looks the same because of some IRL stuff, but I will try to get to it, so it is done before E3.

>>8997

>>8998

>>8999

>>9000

>>9001

>>9004

>>9015

>>9016

Please check here for current best way to update the software:

http://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html

I thought I had a 'how to update from a long time ago' guide there, but I guess not. tl;dr is: Make a backup beforehand. If a 'all in one go, v275->v309' update doesn't work, then try to update in 10-version increments, so go v280->v290->v300->v309. If any of them break, try a smaller increment, like 5, and then 2, until you can get it to work.

>>9018

It probably wrote the path to your log, which is in your install_dir/db folder. I'll check that error reporting code to improve it though!

I already got a bunch of bad files this week and have improved the new thumbnail rendering code for v310, so you may have more luck next week. Please let me know how it goes.

>>9006

It only works for 'file' pages right now. Like this:

https://inkbunny.net/s/1427704

URLs like that should be recognised by the url downloader in the client and be drag-and-droppable.

I still have to overhaul the whole gallery downloader code before I can start adding 'parse a page of thumbs' ability for new sites.

>>9021

Thank you, I did not realise inkbunny could have multiple files per post. Can you post an example URL that is like this? I will update the url class to reflect this. The fundamental ability to pull multiple files from the same URL is the same problem with pixiv manga and twitter posts that I want to fix this week, so I hope I can have this sorted for v310.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

34afb4  No.9026

>>9024

>Can you post an example URL that is like this?

https://inkbunny.net/s/1606666

its the LEAST gay shit i could find for you. hope it helps my dude

>>9024

>I still have to overhaul the whole gallery downloader code before I can start adding 'parse a page of thumbs' ability for new sites.

i understand, i guess hydrus works through the booru sites by making use of the tag system that reins the data organization public and absolute, but in art sites, the tags are more like an extra, they're not obligatory, and don't really have any enforced standards. i can vouch for neither FA nor IB having a standardized tag system.

however, since im not very tech savvy, let me ask you this: how do you do it for the hentaifoundry and deviantart downloaders? they don't really have a standard tag system and really the only common factor linking pictures together is the artists account name/id.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

86cf30  No.9032

>>9024

What you choose to spend your time is on you - but I am curious as to why so much time is spent on developing Hydrus into a scraping program instead of a file organization/tagging program.

IMO, Hydrus should expose a way to write scrapers and then let the community deal with updating scrapers when a website changes or if they want to add a new website. There are a lot of useful organization/tagging features that seem to be on the back burner for parsing/downloading features.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9035

>>9032

because downloading files one at a time is a pain in the ass nigga, hydev is doing the world an immense service. as the amount of media on the internet increases and its production rate accelerates, "folders" just isn't gonna cut it anymore.

at least thats how i see it

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

585ea4  No.9036

I just stumbled here by chance, what does hydrus do that pixivutil doesn't? I'm always up for a better alternative

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

52b89e  No.9038

>>9036

Pixivutil is more geared toward advanced downloading, Hydrus is meant for organizing images with tags on your local system. Hydrus can download from Pixiv, but right now the Pixiv downloader is lacking (at least last time I checked).

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

52b89e  No.9039

>>9036

>>9038

I should note that hydev is trying to fix the biggest problem with the Pixiv downloader by this week, he said in >>9024 that he's working on the downloader for manga posts (right now it only grabs single image posts)

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

c44e27  No.9041

With Microsoft acquiring GitHub, is there any plans to migrate the Hydrus repository to a different provider at some point?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

86cf30  No.9042

>>9035

There are literally hundreds of scrapers out there and it takes <20 minutes to write your own in Python if one doesn't already exist. There are only a few non-shit tagging programs.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9043

>>9041

>Microsoft acquiring GitHub

fuuuuuuuck those kikes, we can't fucking have anything without either microshit or Yidsney buying it these days

>>9042

>everyone can code in every language!

>i want this program to do LESS things!

oh stop it you

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

86cf30  No.9044

>>9043

>"I want this program to do everything poorly and be a bug ridden piece of shit that deletes my database every so often!"

I can misinterpret your argument too.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9045

>>9044

yeah but your misinterpretation is retarded while mine is accurate. using multiple tools becomes a pain.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

0ee470  No.9047

>>9045

>>9044

>>9043

>>9042

>>9035

>>9032

I literally just discovered Hydrus and this board, and this is an interesting discussion! I've always thought about how it would be great to use a tagging system like the boorus on my PC, but I never really sought a solution until now, because I'm lazy. It's also interesting thinking of where things could or should go in the future. What I would hope for, since it's true that content is ever-growing, is perhaps a standard system and database of tags to cover literally all content on the web. Of course, consensus would be needed, so it'd largely be community managed. How this whole thing would work, I'm not entirely sure, but it would make things a lot better. Perhaps AI could even help the tagging along. Again, maybe not a realistic idea, but I'd love if it could one day happen. Ideally, such a system would be an integral part of Windows and the file explorer in the future, and one of the primary ways we access files in the first place, alongside the normal method. I have to wonder though, is this even a discussion being had higher up? Or is it only truly just a niche thing? I would think that in some future where we literally use our mind to control our PCs, the tagged system of organization would be extremely important to efficiency.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9048

>>9047

>What I would hope for, since it's true that content is ever-growing, is perhaps a standard system and database of tags to cover literally all content on the web.

that's what hydrus is though, haven't you read the guide?

you have two public tag repositories, and a local tag repository in your PC.

the PTRs have a guide on proper tagging on the hydrus main site (on hydrus, HELP > LINKS > SITE), essentially it goes something like tag what you see, use namespaces (like character:name, series:name, creator:name, gender:, and so on) and don't add stupid meme tags like pic related.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9049

File: 608aff21b953e40⋯.jpg (8.2 KB, 56x15, 56:15, thicc xddd.JPG)

goddammit 8chan

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

0ee470  No.9050

>>9048

Well, I did only just discover Hydrus, and I've only read the intro that's linked at the top…

Anyway, regardless, what I tried to mean when I said that quoted sentence is a true standard that everyone uses and connects with. Sorry if that wasn't clear. I already knew Hydrus and other platforms have databases that are community managed. I'll happily use Hydrus, but I hope for a standard in the future, which I think may probably only arise if OS vendors implement it and make it a standard way to interact with files, folders, and links (to arbitrary locations).

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9051

>>9050

how do you propose this standard is implemented when every site has its own tags and most sites simply let you type anything as a tag?

what we have is probably the best we can get, contribute to it by not making shitty tags, and replacing shitty tags with the proper ones. You can petition tags to be removed in both PTRs and use "should be namespaced" as a reason for removal.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

3d1d90  No.9053

File: 9fecfb4eb4cbdcf⋯.jpg (429.74 KB, 1400x1000, 7:5, 9fecfb4eb4cbdcfd09dcb4ca43….jpg)

>>9026

Thanks, inkbunny multiple files should all be fixed for tomorrow. I think the files may come randomly just out of order, like 2, 3, 4, 1, due to some unusual markup on their end. Let me know how it goes anyway.

Yeah, my old booru system takes advantage of very common markup across the big booru softwares. For DA and HF etc…, I wrote custom hardcoded solutions (I actually did these before boorus, I think, ages ago), but the whole thing proved way too difficult to maintain on my own. This new system will be much more flexible, allowing any user to create and share and maintain new downloaders for any 'we got pages of thumbs and pages of files' kind of gallery site without me having to fold it into an update.

BTW: While DA and HF tag search works on a more complicated system on their end than how boorus do a direct tag-map, they still spit out standardised chronological and stable results in a grid of thumbs, which is how I am able to do that. On my end, it is all just different ways of going "hey site, here's _search_text_, please give me some thumbs to parse'.

>>9032

>>9035

>>9042

>>9043

>>9044

>>9045

This big downloader overhaul is my attempt to do basically what you are talking about here. Maintaining hardcoded downloading solutions was eating up too much of my time, so I am now moving the whole system to a common user-editable standard. Unfortunately, it turns out that this is a fundamentally more complicated problem to generalise than I thought. I haven't made nice help for this yet because the systems aren't complete, but some advanced users are already contributing parsers and url class definitions to get this stuff going.

I mostly work on what people are interested in, and lots of users are interested in having hydrus able to download en masse from websites. Once this big fucking rewrite is done, I'd like to put my weekly 'big work' time back to better dupe ui, and audio support for videos, and colour searching, and all the other things I've got queued up. I expect to put up a poll for people to vote on what 'big thing' they want most.

This download overhaul has sprawled much more than I expected, and while I have enjoyed it and think it has had good work, I am getting to the point where I am ready for it to be over. I think I originally thought it would take five months, and I'm certain it has been over a year now.

>>9036

Hydrus is foremost an archiving application, for managing files. It works like a booru on your desktop. Privacy and user control are core values. Please check out the help here:

http://hydrusnetwork.github.io/hydrus/

Which has some general screenshots and an intro to the overall philosophy of what I am trying to do here. I recommend prospective users try using hydrus for a few hundred files for a couple of weeks–see if you like it, and let me know if you have any questions.

>>9041

>>9043

I do not know yet. I am thinking about it. Someone mentioned gitlab as an alternative, but I haven't looked into it. Any recommendations or thoughts would be appreciated.

I'm of two minds about the specific purchase–I am no big fan of Microsoft, but I am no big fan of github's current leadership, so if they clean house, it is probably a wash for my personal side of things. I use it mostly as a public host for my releases and code and don't actually do any work or social media through it, so I am not as 'exposed' as other devs may be. But this also means it would be easier for me to move.

But I am thinking about it. I quite like the simple workflow the github app provides, and I've never had any significant problems with the services I actually use, which wouldn't be certain elsewhere. I'm interested to see what Microsoft's business plan is–$7.5 billion is more than 'we provide this to you for free, we are your friend, trust us' brand management money, but they'd also lose the whole userbase if they decided to harshly restrict free accounts, so my current guess is that this is another fucking stupid purchase with no practical non-strategic thought behind it and it'll just slowly decay in relevance over the next ten years like Skype has.

I never made a plan for if my github account gets banned for wrongthink, so maybe if nothing else this is a good kick in the pants to make one.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

0ee470  No.9054

>>9051

Well I haven't really thought deeply about the tagging system itself, all I'm saying is that it would be great if there was a standard that everyone just uses.

But thinking about the tagging system, I think that using AI like I mentioned to help could be interesting in the future. It is already possible to use AI right now to identify with relatively good accuracy what is contained within an image, and it's only going to get better with time.

I also think it could be interesting to let tags themselves have the ability to be tagged, which itself would be inherent in the system I'd like to see happen (you can tag anything). I think it would also be important, if it isn't normal across these systems yet, to have the ability to easily access the properties of a tag, so that you can filter tags themselves beyond using the tags' tags. For instance, I may want to see all tags that count as scenery, but sort by the amount of entries each tag has, or sort by last date modified. The database should also probably be decentralized, to better manage the vast amounts of data that will be generated from everything being tagged. Now I guess this system could reduce the importance of tag accuracy, for better or worse, because you can filter tags themselves by tags. And so you should be able to have default filter settings on so you don't get results you don't want.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

4240ec  No.9055

File: f6033aab2e1dd21⋯.png (99.78 KB, 248x247, 248:247, 1345228229506.png)

>>9054

>tagging the tags

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.


[Return][Go to top][Catalog][Nerve Center][Random][Post a Reply]
Delete Post [ ]
[]
[ / / / / / / / / / / / / / ] [ dir / random / cloveros / cuteboys / erp / fast / hydrus / in / strek / tftg ]