/brg/ - Batoto Replacement General

Previous: What we need from a site
>user accounts
>uploading zips
>reader a la (or just incorporate)
>series page
>limited tags
>search based on series name/author/tags
>comments on series page
>DMs between accounts
>users can follow series, RSS generated of follows automatically updated
>can serve 6+TB of content to tens/hundreds of thousands of users
>reader grouped by language on series page
>scanlation group account/page
>series ratings system

What someone should do
>get with the former admins to find out their goals/backup access
>find someone who has time to actually write shit for code
>probably acquire funding
>like a billion other things

Other urls found in this thread:

ghostbin.com/paste/rrnjg
bato.to/forums/topic/28840-batoto-will-be-closing-permanently/
streamtoken.net/,
alexandria.io/learn
github.com/dloa
ehwiki.org/wiki/Hentai@Home#Free_Archives
twitter.com/NSFWRedditVideo

No comment sections or forums because all that does is encourage faggotry (or if you really need some social crap, at least force user the comments and give a temporary ID like in Sup Forums and /bant/)
Ability to follow/favorite manga and put bookmarks so I can keep going from where I left off.
Ability to organize my favorites list with folders and the like
Let us have the option to read raws.
Give us a good tag system and filtering.

Basically a more visually appealing sadpanda for manga.

who is this qt 3.14

A smart cat

Reposting from last thread
ghostbin.com/paste/rrnjg

Man, I thought for all these years her name was Tamano instead of Tamamo.

>can serve 6+TB of content to tens/hundreds of thousands of users

RIPIP. No amount of techno-wizardry is capable of solving the budgetary server issue.
Not with the level of quality people are asking of it.

>
You're right. I think comments can be useful but when the community becomes /reddit/, everything goes to the trash.

DOA without comments

The only thing I think is halfway feasible would be to make a tracker that just serves .torrent files and magnet links. That way the server wouldn't need to store TBs of content and costs could be kept low.

For that we already have nyaa, really.

>I just hope the next "batoto" can import my follows cause I sure as hell will not manually search for 309 series and check if they got a new chapter everyday.

Sup Forums should get executed on the spot, bunch of spic niggers, normies and retarded teenagers

Everything you listed is quarantined to the big3/DBS generals.

>normies
Get out, normalfag.

yeah just like Sup Forums is quarantined on Sup Forums

Lurk Sup Forums and then go into a DBS general thread. It feels like you stepped into a portal to a fucking hellscape.

We should use hadoop as database because most of data will be unstructured

I just did, the whole board is a hellhole.
Back in 2015 i thought it couldn't get worse but it did.

It's pretty normal right now outside of the generals, but most of the shitposting comes because it's the start of a season.

Anyone rather do peer-to-peer for content if we are just gonna do this from scratch anyways?

Could cook up a simple webserver with Python Flask that acts like a torrent client, quickly grabs the comics from the peer swarm and present to user.

>normalfag
You're fighting a lost battle, user

You're even worse than normies, you're an autistic faggot trying to fit in.

>python
We said no meme languages.

who's tamamo??

These threads seem very dead

The whining which makes up 90% of the posts is on Sup Forums.

The IRC is sorta alive-ish. #manga-dev @ irc.rizon.net

Tamamo no bae

Every language is a meme language, as long as it works it doesn't really matter what to use. Unless it's for optimization purposes, which is too early to think about at this stage.

Don't use a semantic file structure, users aren't browsing the file structure directly.
You want roughly uniform spread of images to directories so no directory ends up with too many images.
Hash images and use as path and nest based of kth character of hash. So n levels gives 36^n directories. Figure out level of nesting desired for max images per directory and total number of images.
Use tables in db to map series/chapter/etc to paths.
Meguca does this ad hoc using english file names because they have users navigating file structure directly, but batoto copy won't need this.

You're going to laugh at me Sup Forums but I'm a scanlator with a 1.6 Mbit speed connection, so my upload speed is like 50 kb/s. You don't actually expect me to seed shit, do you?

If you seed it once, there's plenty of people with a fuck ton of upstream. Even my bashed together scans before compression are 2mb a page, so call it

>I think the bigger issue is having the system work with the sheer number of separate chapters. It wouldn't be viable with a torrent client because they start shitting themselves eventually,
That's correct, this is the whole reason the grand majority of us never used torrents to districute manga. The ones that did usually only did it for whole volumes since it was easier that way, but you obviously can't do that when you're scanlating from magazine scans.

Pic related.

>suddenly we need to think about performance, but fuck it, we did all this work already, were not gonna start from scratch on a not-so-shit-lang
>every (python|ruby|javascript) project ever
never heard "Temporary solutions are neither."?

The idea behind high-level languages like Python is that it's more important to save on programmers' time than CPU cycles. That's why Google uses Python internally extensively, as do a lot of other companies. No one hand-codes a CGI program in C anymore.

Just use Java and get performance and dev time.

They almost certainly use Python extensively for the purposes of business-tertiary applications, like logging and performance analytics.

Anything important is optimized to shit. Programmer's time be damned--that's their whole job description.

A sensible naming scheme would be amazing.

>[Series]_c[NNN]([-MMM])(_vOOO)(_[group and whatever other crap you want to put in here])

Preferably with one series name per series.

i understand the concepts of high level language and programming time, i also know most popular languages (coincidently high level) are shit at performance AND, on top of that, since theyre higher level and easier to learn/use, the programmers that use them are fucking brainlets! thats like putting a person with no hands, trying to hammer down a fucking screw, nothing of value comes out of it (unless you like to do that for shits and giggles)

gonna have to agree with the other user on this.
We should finish building it first (doesn't matter the lang) and then refactor afterwards.

>[Series]_c[NNN]([-MMM])(_vOOO)(_[group and whatever other crap you want to put in here])
whats MMM?
also since youre gonna put so much shit on the file name, might aswell put language

why not just
cd pado.ru
wget bato.to

I would recommend always using Japanese titles in standardised romaji with spaces to denote particles

Would you like a copy of this book then?

its alright, i understand that, but if you plan on refactoring later, maybe even change lang, at least dont waste time on making a solution, make it temporary

If you upload multiple chapters, ie chapters 1-5 of the Faggot Desu series in English by the group "I suck cocks" would be

Faggot_Desu_c001-005_v01_EN_I_suck_cocks_and_include_stupid_shit_in_my_chapter_names_щ(ಠ益ಠщ).zip

NO USER ACCOUNTS

>java
>performance
>dev time
pick one

appreciate it, ill pm the address (y) ;)

Python integrates well with anything that provides a C ABI, so we can rewrite parts of it later if we need to.

makes sense to me, but better put lang and volume number first, for sorting
series_lang_volume_chapter(s)_group.zip

Dumb cat from some mobile gacha

Fuck off autist. The point of this thing is to follow new releases conveniently.

To be honest as long as you don't add forums, avatars, signatures and karma the cancer can be kept in check.

User accounts are required for scanlators at a minimum

Would something like danbooru’s hover text be possible for stuff that has translations but no scanlations?

Would be cool.

>Anything important is optimized to shit.
Is that why YouTube is a massive resource hog nowadays?

YouTube is a massive resource hog because Google keeps trying to push their Polymer web component meme

Names are also a problem, not just the karma and other shit.
By adeding names we tie responsibility to the user rather than to their actions and arguments, which defeats the whole point of anonymity.

what if we ran it on a bunch of networked machines, kinda like the whole Folding@Home thing

"what if we got a billion people to donate their compute power out of the kindness of their hearts"

See all collaborative computing projects ever.
Servers don't run on sentiment.

It's a community thing. You donate your processing power, you get access to the site.
Win/win.

Obviously I should elaborate futher- most people will leech and that should be allowed, but if you made a game (however simple) out of contributing processing power to the site, members would do it if only to compete. If the site got big enough you'd probably have whales actually purchasing servers/etc to host the site for you just to inflate their e-peen.

>be a self-fellating private tracker
nah

The only reason that works for sadpanda is because they have a gay community
Which shitloads of people here are already adamantly against

someone has to pay to keep the lights on
either you tolerate it or you starve

You could also leave those millions of leechers out to dry and only care about Sup Forums's scanlators

Absolutely not.
Give me your tired, your poor,
Your huddled masses yearning to breathe free,
The wretched refuse of your teeming shore.
Send these, the homeless, tempest-tossed, to me:
I lift my lamp beside the golden door.

Why are we replacing Batoto? The site seems fine.

bato.to/forums/topic/28840-batoto-will-be-closing-permanently/

la luz extinguido...

Oh god. That sucks. Checking the thread and I don't see anyone offering much up so far.

Has anyone contacted the staff? Grumpy said in his post:
>Going forward, some of the staff members have and still are in the process of backing up the entire Batoto comic content as they have been notified of this decision well in advanced. They may choose to create a new Batoto or something simpler.

They may already be working on something

...

Ah thanks

If you worry about obscure material not getting enough seed take this problem as a ratio/seeding problem. If you dont want to go the private tracker approach then there are two methods:
1. Use some method to reward seeding obscure material like Stream Token streamtoken.net/, essentially a crypto currency that rewards views, which maybe be tweaked to do the inverse work and reward seeding that which is low seeded.
2. Make seeding a blind operation, somehow seeding randomly, and make the swarm agree on that, just like darknets operate, and moreover how GNUnet is.

I don't understand the reason for closing the whole site. Why doesn't Grumpy just turn over admin to other people?

I mean all that explanation if we are going to for torrents/IPFS. There is a platform called Alexandria you could use as base alexandria.io/learn
Is open source github.com/dloa

Maybe hentai@home got it right, just watch the part on rewards ehwiki.org/wiki/Hentai@Home#Free_Archives
You get to download a free quota after all.

Well he has username/email and password/hash pairs for tons of people.

I can totally see manga@home becoming real.

Why isn't this in the OP?

Are you braindead?

These two understand, starting by making something that 'works' and then worrying about 'optimization' is the first lesson of designing robust software. Ever read Pragmatic Programmer or Unix Way because they all say this in some form?

Lets get started, anyone willing to make the GIT and manage pull requests? Can just stick it with a temp name for now. Seems like we have a lot of concern about naming conventions and content hosting that we should nail down.

I'd count using a P2P solution as early optimization (trying to minimize server costs). Go with a standard relational database that stores relative paths to the chapters or series. Use the MU tag system. Use a standard JS comic viewer. Make a proof of concept and show it here.

ewhal4 (nyaa pantsu owner) has already said he's interested in implementing a normal manga reader. No IPFS memes. He has a git repo and there is also one made by the IRC people for the IPFS meme project.

You mean no P2P solution but a centralized server?

>ewhal4 (nyaa pantsu owner) has already said he's interested in implementing a normal manga reader. No IPFS memes.
>No IPFS memes.
at least someone on Sup Forums is rational

I dont want to be rude, but yet another centralized website is ought to be taken down. Instead of accusing decentralization of being a meme think about that first.

I agree. To kind of counteract that, how about a centralized site that also offers a torrent magnet link for each chapter on the site, and make weekly bundles of all the magnet links on the site, like what Wikipedia does with their database dumps? That way, even if the site goes down, the content that other people decided to seed would still be available.

Nobody would seed anything ever.
Okay, maybe the latest chapter of one piece.

Sure, that is the kind of thing that could save us at the end. Is all about countermeasures.

It is an alternative, doing nothing is the worst choice.

>but yet another centralized website is ought to be taken down
Make content backups public. DJT does that with their cornucopia of resources and it works fine. The current maintainer gets bored and someone else takes the backup and keeps it alive.

This daiz naming scheme is fairly good but it has some stupid problems
The (_v000) tag is "supposed" to indicate the source of the raws (v01 for tankoubon volume 1, mag for magazine, etc), but probaly 50% of users just think it's a tag for which volume the chapter is from and use it even for magazine/webrips

It needs to have separate places for a.) showing which volume the chapter "belongs" to, and b.) showing whether it is a tankoubon scan

Centralization isn't a problem if there's public backups or a team dedicated to it.
Also thinking a large amount of people are going to be willing to seed a dozen TBs for p2p to work is a pipedream.

What about having "user classes" where people get higher titles for a larger seed size, and this allows them to access special forums for their user class?

Also, what do you guys think about ip-limiting the max number of daily chapters a user can download, and exponentially increase this limit for people who seed a lot and have a higher user class? IMO that's a real benefit for someone seeding a bunch of stuff.

>sort by group before chapter
This is incredibly bad for usability
if you put them all in one folder you get
series_agroup_c12
series_bgroub_c1-10
series_cgroup_c25
series_dgroup_c11
series_dgroup_c13-24

So, a private tracker. Fuck off, that won't fly as a replacement for Batoto, just yet another obscure circlejerk..

But thinking a team dedicated to store backups can't being taken down is a pipedream!