habahut

1e87ba20e81df142b2f93e0e6c5992d677572fec

habahut

Install vim: must have +python and +signs 

install: http://www.vim.org/scripts/script.php?script_id=2508

run these commands:
sudo apt-get install php5-dev php-pear
sudo pecl install xdebug

grab location of xdebug.so from installation output, put here: /etc/php5/apache2/php.ini
zend_extension=`/path/to/xdebug.so`

habahut

nnoremap <leader>D :DiffSaved<cr>

function! s:DiffWithSaved()
    let filetype=&ft
    diffthis
    vnew | r # | normal! 1Gdd
    diffthis
    exe "setlocal bt=nofile bh=wipe nobl noswf ro ft=" . filetype
endfunction
com! DiffSaved call s:DiffWithSaved()

habahut

It is likely that you will need to update your copy of php running on the devbox. To do this,
 you will need to navigate to the manual link to download the newest version of PHP in the
 placelocal repo. Browsing to the download link, and copy and paste it like so:

`cd /vagrant`
`wget <insert download link here>`

This will download the version to a file in the current folder. Type `ls` to find it, then:

`tar -xzf <the file name>`

habahut

:GrepTab <search term> <path> will open a new tab and show you all results from ack-grep of the search term on the path given. No path will default to current directory  
(I recommend you ad `set autochdir` to your .vimrc as well, so that the "present working directory" is always at the file your editing. It makes searching for stuff easier)

   nnoremap <leader>f :GrepTab "example binding

   function! s:NewTabGrep(...)
       let args=split(a:1)
       if len(args) == 2
           let dir=args[1]
       else

habahut

My understanding so far:
one object per topic: for example TPS could be called the TPSTracker
Each object will have many little processes it is forwarding data to, so for example in TPS there would be a little process for "# of jobs started", "# of jobs on each TPS server" etc...
the TPSTracker will forward each kafka message it recevies to each of hte little entities, which will maintain their own state on the aggregate totals for their piece, throwing away the rest. So the TPS load piece would receive the entire TPS message from kafka, and only update the state for the load on the TPS servers, ignoring everything else. 
Then, we stringify the state of each individual peice and send that to the client, and allow them to put that data into the charts on the clients page.



btw Kyle your idea of not bothering to verify producer output makes a lot more sense when we look at it this way, though I still think we should enforce some regularity :)