Reasonable Fidelity

What?

The idea behind this app is to record whatever actions you would when testing a
web app, then replay them later and compare their level of fidelity. Attributes/files you do not wish to compare can be added to 'ignore' files.

How?

• Open your favorite browswer (let's pretend it's Chrome)
• Open the developer console
• Open the 'Network' tab and select 'Preserve logs'
• Interact with application
• Right click the developer console, select 'Save as HAR with content'
• Use this file to replay and compare
• Fine tune ignored attributes in config/ directory

Some Features

• CSV or full text output
• Multiple configuration environments
• Cookies or dedicated token
• Plugins (written in Python)
• ...

TODO

• more advanced attribute extractors
• more configurable login
• Point Free composition!

Using

Valid Arguments:

help:
    display help

sanity:
    check requirements & help explain issues

run:
    run tests
    options:
        --csv: output csv data rather than plain text
         -source <file_name>: specify alternate .har file
         -config <dir_name>: specify alternate config directory
         -forcelogin: attempt log in to service rather than trust cookies/token (experimental!)

modified:
    show which lines were changed from the reference capture

timestamp <startedDateTime> <time>:
    return a timestamp for 'start_at'/'stop_at' configuration

edit <filename>:
    edit har file -- for now, mark actions for deletion

reset:
    re-create default configuration file

Building

From scratch, really...

OS X

brew install opam

Linux

sudo apt install m4
wget https://raw.github.com/ocaml/opam/master/shell/opam_installer.sh -O - | sh -s /usr/local/bin

Then (Common)

opam init
opam update && opam switch 4.09.0

To install dependencies .opam file from dune-project:

dune build @install

Curses

The built-in editor relies on ncurses (v5). On some Linux distros, you may have to run make prepare_curses

More building...

Same goes for lymp that is running cold that's a bit too... dated.

Run:

make prepare

When everything is built properly, we will want to create symlinks to clean stubs:

sudo bash -c 'for lib in ~/.opam/default/lib/stublibs/*; do libname="$(basename $lib)"; ln -s $lib /usr/local/lib/${libname/dll/lib}; done'

Specifics of this code

My first ReasonML app! So, yeah, it's not perfect,

Things to keep in mind if this is your first use of Lwt promises

• Lwt is our Promises library
• Lwt comes with infix operators:
• >>= equivalent to Lwt.bind
• >|= equivalent to Lwd.map (right side does not return a promise)
• '=<<and=|<` are reversed versions

Misc Notes

Parallel queries

• We can use Lwt.run(Lwt.join(...

• execute.get is good

• we could run up to n threads in parallel

• ...or until we meet execute.post

• All pending threads to be stored in a list...

• Orchestrated in execute.get?

• Q: what to do about pause when > 0.?

• Do something about non-csv output, which is broken up

• {action, url, query_string, post_data)