3 ms·

How would you develop an application to do offline routing like my GraphHopper road routing project? Such applications have to use the RAM or disc access provi

How would you develop an application to do offline routing like my GraphHopper road routing project? Such applications have to use the RAM or disc access provided by the OS. A simple cache will not do it. Although I could imagine that JavaScript is fast enough these days you still need access to lots of memory and load several hundred MB from somewhere/disc

doublec 13y ago
You could download the routing data into IndexedDB and use that for your file system.
- karussell 13y ago
  And you know how fast that would be? Using an ordinary DB for graph algorithms is already slowish on a powerful server ...
  - doublec 13y ago
    IndexedDB allows direct access to the blobs like a file: https://developer.mozilla.org/en-US/docs/Web/API/FileReader https://developer.mozilla.org/en-US/docs/Web/API/FileReader
    - karussell 13y ago
      Intersting! (btw: this is only read only but useful)
      - doublec 13y ago
        There's a writable interface to IndexedDB files as well: https://developer.mozilla.org/en-US/docs/WebAPI/FileHandle https://developer.mozilla.org/en-US/docs/WebAPI/FileHandle
Yoric 13y ago
Interesting question. Bulk loading/saving big binary data from/to IndexedDB is basically as fast as accessing the file system. Using TypedArray or (upcoming) BinaryData, JavaScript can also manipulate binary data pretty quickly. I haven't attempted doing any serious graph processing in JavaScript, but these days, JavaScript is extremely fast at tree manipulation (much faster than OCaml, which has long been the fastest high-level language for this kind of thing), so I would imagine that it's quite suited for the task. If you are serious about this, you should come and chat on irc.mozilla.org, channel #openwebapps for instance. We'll be happy to help you get started, plus that kind of discussion is often how new features and optimizations get started.

doublec 13y ago

You could download the routing data into IndexedDB and use that for your file system.

karussell 13y ago
And you know how fast that would be? Using an ordinary DB for graph algorithms is already slowish on a powerful server ...
- doublec 13y ago
  IndexedDB allows direct access to the blobs like a file: https://developer.mozilla.org/en-US/docs/Web/API/FileReader https://developer.mozilla.org/en-US/docs/Web/API/FileReader
  - karussell 13y ago
    Intersting! (btw: this is only read only but useful)
    - doublec 13y ago
      There's a writable interface to IndexedDB files as well: https://developer.mozilla.org/en-US/docs/WebAPI/FileHandle https://developer.mozilla.org/en-US/docs/WebAPI/FileHandle

Yoric 13y ago

Interesting question. Bulk loading/saving big binary data from/to IndexedDB is basically as fast as accessing the file system. Using TypedArray or (upcoming) BinaryData, JavaScript can also manipulate binary data pretty quickly. I haven't attempted doing any serious graph processing in JavaScript, but these days, JavaScript is extremely fast at tree manipulation (much faster than OCaml, which has long been the fastest high-level language for this kind of thing), so I would imagine that it's quite suited for the task. If you are serious about this, you should come and chat on irc.mozilla.org, channel #openwebapps for instance. We'll be happy to help you get started, plus that kind of discussion is often how new features and optimizations get started.