New filename format?

I noticed the title bar now appears in the filename. Is there a logical reason for this, or was it just added in?

It's a new feature - probably helps for google indexing / people who download images.

However, since windows has a hard limit of 255 character for filenames (full path, including folders) this could raise a new issue.
Unsure how the server handles filenames of 300+ characters like absolutely_everyone

Edit; derp, it's already discussed in topic #9127/p12

I've run into the filename limit problem with downloads from Derpibooru, which adds the ten tags with the highest postcount to the filename. Normally it isn't an issue, but with tags like "this will end in tears and/or death", not to mention it names the punctuation...

That's Derpibooru's problem, not Danbooru's, and it doesn't use the same engine as Danbooru.

It may help Google but I am fully non-plussed by this as it is now harder to keep track of unique images and it is too much in terms of character limitations on most folks computers, and WILL necessitate the purchase (at the user's expense) of a whole new device with the limits removed, an expense that is fast becoming not an option in these times. A simple ID number is the quickest and easiest to keep track of FOR A REASON and is the way most big firms track and trace inventory and this new system reminds me of the early days of postal service, woefully inefficient and highly subjective. A real down vote from me today.

RJTAYLER said:

It may help Google but I am fully non-plussed by this as it is now harder to keep track of unique images and it is too much in terms of character limitations on most folks computers, and WILL necessitate the purchase (at the user's expense) of a whole new device with the limits removed, an expense that is fast becoming not an option in these times. A simple ID number is the quickest and easiest to keep track of FOR A REASON and is the way most big firms track and trace inventory and this new system reminds me of the early days of postal service, woefully inefficient and highly subjective. A real down vote from me today.

You can already turn this off in your user settings, you know. Also, purchasing whole new device for longer filenames? What?

This is also a disaster for Unix users since, length and non-uniqueness and redundancy with users' own classification systems and making it harder to look an image up all to one side, leading hyphens are going to screw up a lot of scripts and commandline tools.

Yes, there's an option (buried at the bottom of 'advanced options'), so at least we can turn it off, but I wonder how many new users will ever know it's a possibility? If SEO is such a priority, perhaps the bad filenames could be served only to non-logged-in users (which would include the search engine bots) rather than users with accounts.

Cyprianus said:

This is also a disaster for Unix users since, length and non-uniqueness and redundancy with users' own classification systems and making it harder to look an image up all to one side, leading hyphens are going to screw up a lot of scripts and commandline tools.

What nonuniqueness? Md5 didn't go anywhere, it's still at the end of filename. Leading hyphens can indeed lead to issues, though, they should get removed I guess.

Also, you should move your suggestions to issues topic or github, albert rarely monitors other forum and decision is up to him.

Works fine on my Mint and Opensuse Linux systems.

Please, I request that this ''feature'' should be removed. I'ts kinda useless and unnecessary, specially for the users that download pictures regularly, because they will have to check which image is not repeated in their folders, at least in Google Chrome.

... Is there at least an option to disable this ''feature''?

(Sorry for my bad english)

EDIT: Nevermind, i found the disable option. Anyway, think about people who doesn't have an user account.

thanks to the guy who pointed out about the option,didnt know about it, maybe this topic should be moved up somehow so people can see it (plus c'mon, you can do an acount in what, 5 minutes?)

MrVargasProductions said:

Please, I request that this ''feature'' should be removed. I'ts kinda useless and unnecessary, specially for the users that download pictures regularly, because they will have to check which image is not repeated in their folders, at least in Google Chrome.

... Is there at least an option to disable this ''feature''?

(Sorry for my bad english)

EDIT: Nevermind, i found the disable option. Anyway, think about people who doesn't have an user account.

Why on earth would they have to check ? When you save, and a file with the same name in that folder is found, the browser asks if you want to overwrite. You are always given a choice.

This excellent innovation in itself guarantees there will never be two files with the same name on Danbooru server --- unlike the old inferior naming system.

But the old filenames are just the md5s of the pictures themselves. What are the chances of two uploads here having the same md5 hashs?

I certainly won't look for it, but there was some discussion here ( in which I did not participate ) of eventual duplicated sets having to happen.

Still the best part of the new system is that artists have their names joined to their works, without error: everyone with a copy can immediately see who painted it.

MD5 produces a 128-bit (16-byte) output. Thus, there are 2^128 ≈ 3.40e38 possible MD5 hashes. The expected value for the first collision occurs at the square root of the sample space, that is, 2^64 ≈ 1.84e19, or 18.4 quintillion images.

MD5 is no longer secure, and differing files can be intentionally formed with the same MD5 checksum. However, it would be very difficult, if not practically impossible, to manipulate the data in this manner and still produce a valid image format as well.

Given that we're shy of 2.4 million, the probability of a collision can be estimated as n^2 / 2N.
2,400,000^2 / 2^129 ≈ 8.46e-27.
Less than 1 in 100 septillion.

Hypothetically possible mischief aside, the odds are astronomically small.

In case the collision happens, the image cannot be uploaded, because Danbooru assumes md5 hashes to be unique.

There was at least one collision already, happened to me years ago. The upload was not good enough to be worth the hassle so I just dumped it.

Cyprianus said:

...leading hyphens are going to screw up a lot of scripts and commandline tools.

Caught me in the end. Moving files etc. with Rsync or the file manager, no problem. But...

.

I accumulate enough images in a directory, say D14, until that directory is really heavy and slow, say 10GB; then I move that directory to storage on an external file, and start another, say D15.

To have the same structure of folders, I've always used the terminal ( non-root ):

*find * -type d -exec mkdir /where/you/want/\{\} \;*

adapted. And then copied the result back to where the original main directory was.

.

This time due to the leading hyphens it stopped, throwing up errors.

Anyone wanting to correct this ( and it won't happen again ) for this issue, will find pyRenamer safe:

Open the major directory there; under options, select files_and_directories and keep extensions; and go to Substitutions below. Tick replace and enter, -- and _

You can put anything legal in the second instead of _ ---- just not leave it empty. Hit Preview and then Rename. Takes less than 2 minutes for 10GB...

I use a similar GUI tool in Windows called FastFileRenamer, though it can have the second field blank to remove portions. Great for condensing the massive filenames people put on videos these days.

OOZ662 said:

I use a similar GUI tool in Windows called FastFileRenamer, though it can have the second field blank to remove portions. Great for condensing the massive filenames people put on videos these days.

@OOZ662

A good alternative is ReNamer by Denis Kozlov. It's still being maintained by the developer (latest release 11 June 2016).