2.5 Years to put a webpage on a tablet

A Journey into the depths of the Android Platform

Transcript

1. How  it  took  us  2.5  years  to  put   HTML  content  on  a  tablet     -­‐  A  journey  into  the  depths  of   Android  
2. a  confession…  
3. We  need  to  be  able  to  play   websites  on  tablets  
5. Oh,  and  it’s  got  to  work   without  the  internet!  
6. Ok,  great,    but…  it’s  not   secure?!  
7. “There’s  no  way  we  want  to  take   responsibility  for  the  Android  OS  -­‐   it’ll  be  a  rabbit  hole”   -­‐  past  me  
9. The  no  touch,  touch  device  
10. Don’t  let  me  out,     don’t  prompt  me  for  anything       No  navigaGon  bar   No  status  bar   No  popups   No  noGficaGons   No  toasts  
12. Get  those  updates  out!  
13. Now  we  need  it  to  work  on   all  these  other  (x86)  devices  
14. Community  driven,  but  one   primary  maintainer   Chih-­‐Wei  Huang   Run  by  Intel.    Opensource,   but  not  community  focussed   Android-­‐x86.org   01.org/android-­‐ia  
15. Hey,  what  the….     my  awesome  cat     video  wont  play!  
16. Don’t.  Fail.  Ever.  
18. ANR!      OMG!  
19. Zy-­‐what?!  
20. Zygote  
22. Zygote  
23. Watchdogs.  
24. Hey,  why  aren’t  you   updaRng  Screens?  
25. System  Images   •  Kernel   •  Exactly  what  is  says  –  it’s  the  kernel   •  Ramdisk.img   •  Contains  init.   •  System.img   •  -­‐  ‘everything  else’  –  the  Android  layer  
26. I  want  more     (remote)  control!  
27. Argghhhh!  Everbody’s   talking  about  beacons!  
28. Other  Stuff   •  Proxy  networks   •  Android  STILL  doesn’t  support   authenRcated  HTTP  proxies  –  so  we  built   it  (twice)   •  Scheduling  naRve  apps   •  Peripherals   •  Sonar  sensors  /  Lap  Counters  /  USB  lights   •  AllJoyn  Support  
30. Where  has  a  webpage  on  a  tablet   taken  us?    
31. Lessons  Learned   •  Use  the  code  as  a  reference,  not  the  docs     •  (many  hidden  methods  etc)   •  Mission  criGcal  code  should  not  be  in  the  ApplicaGon  layer!   •  Move  it  to  the  naGve  layer.   •  Hardware  manufacturers  have  a  penchant  for  inflicGng  pain   •  They  don’t  like  making  things  easy!   •  Consumer  devices  expect  user  input  /  have  many  quirks   •  People  have  already  thought  about  this  stuff!   •           hardware  watchdogs  etc.   •  Google  like  to  do  things  *almost*  the  same  but  not  quite   •  Linux  kernel   •  bionic  -­‐  their  version  of  c  library   •  custom  init  service   •  bluedroid  
32. Show  them  a  demo!  
33. “Turning consumer devices into enterprise networks”
34. Thank  you!   Nick  Marfleet   @nickmarfleet   theworkshop.screeneros.com     www.screeneros.com   www.blocksglobal.com