Discussion
I made a little web tool to split 8-page foldable zines into their corresponding page images:
https://ragman.net/zinecutter/
I wrote it in a bit of a fugue, and only just got it working well enough to show it off. Any feedback people have is appreciated! :D
Thanks to @robb for the inspiration!
Permalink: https://www.ragman.net/notes/57
I can't be the only writer seeing this: my sales have tanked over the past few months. I have 9 novels & a short story collection out in the world, widely available in all venues. Up until early this year, a few books would sell every single day. With little to no promotion.
The biggest factor I can think of is the rise of AI slop for sale. Maybe it's also partly because of the economy & lack of money for non-essentials.
But, wow, is it demoralizing.
become ungovernable
🚨 Carney is saying in French the USA tried to limit
“Discoverability online of Canadian and French culture and language!? Canadian products”
Incredible!!
Tell me again that Digital Sovereignty isn’t absolutely critical!! Hello Canadian Fediverse and also @Paulatics
Mesmerized by Dennis Hlynsky's overlapped timelapses of starlings, crows, fish…https://vimeo.com/dennishlynsky/videos/sort:plays/format:thumbnail
*pulls string*
"The black-crowned night heron says, 'Squawk, plup, woc-a-woc, rattle rattle.'"
@SRAZKVT i'm sorry for getting upset at you earlier. that was my fault. i agree with you here. 
i get very upset at linus for other matters and it affects my judgement.
note that i'm particularly looking to do the opposite of DirHash, which actively obfuscates content and which google wants to inflict upon pypi. one thing zip files do really well is having a plaintext metadata table, and it makes them more useful than tarballs for source releases since they essentially have a built-in manifest.
i also think that it could be possible to generate an entire hash computation tree which is used as the basis of comparison instead of just the top hash.
if git already does this: cool! but when pants used an object database, directories were completely opaque and encoded as serialized protobuf blobs, much like DirHash:
More precisely, the hashed summary contains a single line for each file in the list, ordered by
slices.Sortapplied to the file names, where each line consists of the hexadecimal SHA-256 hash of the file content, two spaces (U+0020), the file name, and a newline (U+000A).
one thing that an explicit hash tree achieves is greater collision resistance. instead of modifying a single file (or adding extra space in a .tar.zst) to achieve a collision over a whole tarball, you really need to modify an individual file to get an important collision. which brings me to...
one thing pants actually did right with its annoying little object store interface was to record the checksum and the length for all files, which makes it much more difficult for an attacker to add junk data in an unevaluated region (e.g. within a /* ... */ comment) to achieve a hash collision after modifying a file to do evil. a length is another kind of checksum, and it supports the integrity of the standard checksum.
there is also the option of retaining hash computation trees within a file, if using a tree-like hash such as BLAKE3, and this would generally take O(log2 n) space with n the number of chunks in the file. but the way files change can be very strange (e.g. insertion at start breaks chunking, unlike directory trees), so i don't think this would be useful. we should probably just stick with hash + length.
@SRAZKVT you of course were way ahead of me when you introduced incremental identifiers just above.
i like the idea of calling these "inodes". i think they should be local to a representation of a directory tree, and possibly namespaced separately between files and directories.
you're right that we would need to have a reverse index mapping checksums to inodes. i actually think that's ok, because i want to explicitly check for whether the object database has this checksum yet, and insert it if not. i feel that allocating an inode is a useful way to mark this operation.
i also feel that operations upon inodes (like constructing views and/or overlays) are generally a different matter than checksums, so i feel this separation makes sense.
finally, if we generate a hash tree representation to send over the network, i feel it might make sense to also assign it similar inodes if only to make the point that it's a kind of database in itself.
if we use something like SHA-3 or BLAKE3 for file hashing, our potential output space is variable, but an inode will need to be recycled at some point.
sharding into separate file and directory inode namespaces is probably a good idea, and perhaps we furthermore shard by checksum variant (since directory checksums are likely to be a subject of research, this is probably a good idea).
however, at some point, we might overflow. i suppose we could simply add the number of bits we take from the checksum + the number of bits for the file length together. we also don't need to do much more than increment the inode--it doesn't need to be a number-like object.
so perhaps in the protocol we specify that an inode is a k-bit string and we can just increment it and abort loudly if it overflows!
@catsalad I love this!
@Ashedryden I’m not sure, but I know what you mean. I’m really good at ignoring it, but if I think about it, it’s over.
@ramsey exactly
Today has been AWKWARD but at least I have new yoshi game to focus on and trash tv for background noise
i wonder if there's an nginx plugin for this that lets you make/revoke ephemeral share URLs. like i know i can just do this in 5 minutes with a python web server, but i also love just having a website that is just "files in a directory" where the web server part is basically invisible in the OS infra
@jonny I've been writing a little P2P tool called `sned` for quickly sending files between devices.
Just yesterday I added the ability to forward TCP connections through it via a local TCP port. (My idea was for a p2p ngrok alternative.)
So, you could start up a local server, have someone run `sned meet` to configure aliases for each other. Then `sned tcp-connect` them to a web server you're running locally.
https://codeberg.org/nfnitloop/sned/commit/adae1dfdcdf6c4653817ea43c94ca0b1e6384c2a
Because Omarchy news is going around, not so gentle reminder that the author is a vile, miserable, racist asshole who is calling for the eradication of entire peoples.
@heidilifeldman BARF?
@bluegreenandfree 🤷🏻♀️
'type "shut up" to end your session' is top UX
RE: https://bsky.app/profile/did:plc:73dpznbu4wqwtcyurwbiulov/post/3mtog3j3etf23
You know how I know the right wing claim that "Christianity is under attack" because NJ Gov Sherrill won't let MAGAs harass people outside abortion clinics—is actually a front for white supremacy?
Right now the City Hall of Ocean City, MD is suing a Church for the "crime" of turning their church into a homeless shelter—and I have not heard one single conservative Christian pastor condemn this Govt overreach punishing a Church for following the teachings of Christ.
Not One. And they won't—ever.
Is there a .conf TLD? I feel like we need a .conf TLD.
surely nothing will go wrong with a .conf TLD.
Do not fucking create deepfakes even to mock Trump and Republicans. They won't care and it only exacerbates an environment of disinformation.
A wee find from my walk this afternoon. I think I surprised them 😮
With the right magical primer, it is possible to summon the Devil from the bowels of the Earth.
#Germany #folktale #folklore #Devil
https://wiki.sunkencastles.com/wiki/The_Devil_in_the_Shaft
Sorry, the app encountered an unexpected error
Install bonfire.cafe
Get the full app experience
