Adrian Cochrane @alcinnz@floss.social

@alcinnz @codesections @rain sorry, you're both wrong.

There's the String and the str (and several other string types). String is heap allocated; it stores a pointer, a length and a capacity (actually it wraps a Vec<u8> which stores these three). str, a string slice, is not allocated anywhere; it's pointing into somebody else's memory — maybe into a memory that belongs to a String, maybe into a stack buffer, maybe into an mmaped file, and maybe into .rodata

@codesections I think it's nice for languages like Haskell.

The low hanging fruit here would be to chart the pre-existing "browser history", but that doesn't offer insight into real usage and interaction with a webpage.

What's type of strings do you think is best? Why?

(related: https://www.joelonsoftware.com/2001/12/11/back-to-basics/)

@em for those interested in the entire article, you can find that here: https://mattstoller.substack.com/p/the-slow-death-of-hollywood

Websites track browsing habits of their users. But what if the users wants to track their own behavior?

Are there browser plugins that allow users to build their own web usage profiles without that data leaving their browser/computer/control?

@bugaevc @jeffalyanak @kungtotte

> when C was made you counted system memory in bytes, so the C-style strings added just one byte of overheard

> hm, indeed, one byte overhead (vs two bytes for the length) could be a slight advantage.

I could be confused, but I don't *think* rust's overhead is just an extra byte. If you only have a 1 byte length, you'd be limited to 256 chars, right? I think rust strings use 8 bytes for length & another 8 for capacity

@anne @riking @KitRedgrave I'm so happy to see that I'm not the only one who's sad to see network transparency go 😁

Here's a good paper analyzing network performance of X in practice: https://keithp.com/~keithp/talks/usenix2003/

There's tons of low-hanging fruit in X network performance. For extreme examples, see https://lists.freedesktop.org/archives/xcb/2010-June/006111.html and https://bugs.freedesktop.org/show_bug.cgi?id=4232

We hoped toolkits and apps would start using XCB to take advantage of the good design of the actual X protocol, but as far as I can tell, people had already given up on using X over the network by the time XCB was ready, so there was little pressure to make good use of it.

Network transparency could have enabled very cool migration and replication stuff that still nobody can do today, certainly not with VNC-style remoting. But apparently the only thing that ever did it was Emacs: https://www.usenix.org/legacy/publications/library/proceedings/usenix02/tech/freenix/full_papers/gettys/gettys_html/index.html

So many cool ideas didn't catch on, and it's sad.

@codesections @jeffalyanak @kungtotte Rust uses usize for length; that's 8 bytes on 64-bit systems. C was created on a 16-bit system, I believe; that means the pointer and the length take up two bytes each.

@grainloom I think compatible and competing implementations of Nix would be a good thing! I don't like writing anything new in C in 2019, but I hate it less than C++, so I could be okay with that. 😁

I just got around to updating my movie recommendations: https://adrian.geek.nz/movies.html

I stopped forwarding the Mubi recommendation, it wasn't what I thought it was.

Instead I am pleased to recommend the audio comedy Magus Elgar! https://maguselgar.com/)

As I wrote there: 'Filled with energy and expressive voice acting, the members of "Magical Anomoly Interdimensional Locators" have a really humerous dynamic as they seek "Scientific Tools Augmented with Magical Powers".'

@bugaevc @codesections @alcinnz In most assembly flavors it’s very easy to loop over a string and have the loop stop on null. It made for very fast and efficient compiled code.

@yulefyre The ad industry had their chance, but they decided to ignore Do Not Track (DNT), so fuck them. ¯\_(ツ)_/¯

criminalize DRM