maverick1995 / webpagetest Goto Github PK

Azure provides locations that largely do not overlap with EC2 so adding support 
for running on Azure will open up more geographic locations for testing.

- use the remote web inspector interfaces to get the network timeline data and 
events

Add support for tracking socket-level connection errors to wptdriver.  Right 
now only requests that make it far enough to send the http request out are 
being tracked.

Create a cross-browser bookmarklet for testing WPT script commands locally.

Build a relay that can broker test requests from WPT to test agents: 
https://sites.google.com/a/webpagetest.org/docs/system-design/webpagetest-relay

Right now when you export a HAR file it's always called "pagetest.har". The 
user has to change the name *every time* in order to differentiate downloads. 

It would be better if the filename was more specific to the actual site being 
studied. I would suggest URL or URL+ID. For example, the HAR file for 
http://httparchive.webpagetest.org/result/110315_6W_15YH/3/details/ could be 
"www.backup.com.har" or "www.backup.com.110315_6W_15YH.har".

Allow for pcap files to be uploaded and automate converting them to HAR's

When the agents capture video, they should include the timestamp of the last 
visual change as a metric.  Right now it is calculated on the server based on 
the video frames.

- Implement the more advanced features in the firefox agent

After we support HAR files natively, add support for importing existing HAR 
files through the UI and API.

Generate a document that explains what is done right and what can be improved 
based on a webpagetest run.  The report should be comprehendible to management, 
so that someone improving performance can easily show the result of their work 
to their boss.

Deciding what to say in this report is tricky.  Might test by showing a report 
to a project manager who understands the web, but is not a coder. 

Here is sample code for serializing the DOM to json in a format that Page Speed 
understands: 
http://code.google.com/p/page-speed/source/browse/chromium_extension/trunk/src/p
agespeed_chromium/extension_files/pagespeed.js#741

What steps will reproduce the problem?
1. Use the web api to run a job and use a picky XML parser on the response
2. Or, try to use the visual comparison tool with an XML parser that is anal 
about &'s in its input


What is the expected output? What do you see instead?

  XML with properly escaped elements, eg, "&" turned into "&"


What version of the product are you using? On what operating system?

  A private instance of 2.3. Problem is still present in trunk (lines 385 and 413 of runtest.php).

  O/S is not relevant, though the simplexml in my PHP interpreter seems to be picky about what it will parse.


Please provide any additional information below.

  In particular, the ampersands within the URLs of the two places that detailCSV elements are produced need to be escaped. Some XML parsers choke on the entire XML document without proper & escapes in them, and all XML parsers should understand how to unescape them.

Example of said choking (unfortunately, I've since deleted the test xml and php 
files). It clearly shows the parse error and the element it barfed on.

[Sun Sep 25 11:28:29 2011] [error] [client 207.162.200.253] PHP Warning:  
simplexml_load_file(): test.xml:11: parser error : EntityRef: expecting ';' in 
/data/wpt.flirble.org/testxml.php on line 3
[Sun Sep 25 11:28:29 2011] [error] [client 207.162.200.253] PHP Warning:  
simplexml_load_file(): 
<detailCSV>http://wpt.flirble.org/csv.php?test=110925_9M_Y&requests=1</detailCSV
 in /data/wpt.flirble.org/testxml.php on line 3
[Sun Sep 25 11:28:29 2011] [error] [client 207.162.200.253] PHP Warning:  
simplexml_load_file():                                                          
          ^ in /data/wpt.flirble.org/testxml.php on line 3

- tables comparing various metrics
- bar charts for the metrics
- overlaid waterfalls with adjustable transparancy

Finish the system design/architecture docs: 
https://sites.google.com/a/webpagetest.org/docs/system-design

Investigate options for integrating with web driver for extracting performance 
data from existing automated testing (particularly continuous integration 
situations)

This might be as simple as enhancing the existing browser support in web driver 
to "reserve" a machine on WPT and remotely connect to it (after it has been 
started up with the hook injected for instrumentation).

Migrate from the tab-delimited results files to HAR files for the test results 
(with all data from all of the runs in a single compressed HAR file).

This will significantly reduce the number of files for storing the results and 
should also reduce the disk space required.

Once we can natively operate on HAR files then importing them will also be 
trivial.

This will require extending the HAR spec to support the data that we capture 
that is not already part of the spec.

From the perspective of the code for loading the results, the loader routines 
are already centralized so we can make them capable of dealing with both file 
types.

Since the lock files are acquired frequently, add support for a temp folder 
setting (which would let the site admin put them in tmpfs).

Hi,

Can you please add the date and time of when the test took place to the Summary 
page, top left? I now have to click through to the waterfall chart page and 
find a date/time there :(

E.g. 

Web Page Performance Test for
www.pitchup.com/search

From: Gloucester, UK - IE7 - DSL
Date: Mon, 28 Feb 2011 19:18:14 GMT

Increase the number of simultaneous videos from 9 to 25

We already ignore a margin around the page for start render, we should extend 
the logic to do the same for comparing video frames so we don't count status 
messages against the visual complete time (and update video just for scroll bar 
changes)

Add support for decoding chunked encoded bodies

Currently, wptdriver can be configured to run either Chrome or Firefox, but not 
both on the same machine.

Add the capability of running tests for different browsers from the test 
machine.

Add support for decoding SPDY streams into individual requests.  Optimally this 
will also include enhancing the connection view to support concurrent streams 
on a given connection (possibly using connection.stream for the connection ID's)

Firefox draws the browser content and toolbar/chrome all as a single window so 
there is no separate window for just the html content.  Need to figure out a 
way to identify just the browser part of the window and use that for screen 
captures and render detection.

Add support for IPV6 sockets/addresses (wpthook and php server)

Firefox, Chrome (IE9?)

Content blocking does not appear to be working any more in wptdriver

Add support for configuring wptdriver EC2 agents through user data (server, 
location and key).

Optimally we could also have Chrome and Firefox both auto-update when the 
instance starts up so the browser will always be up to date.

Suppose I ran webpagetest a month ago, and run it again today.  I want to see 
what changed, so that I can fix any regressions.

Add support for using Amazon's Queuing service for the work queues (for 
supporting larger scale testing).

Migrate the legacy IE agent to the new agent platform

As applications get more complicated, and as optimizations take more advantage 
of the available technologies, scripts play a bigger and more fragile role. 

Measuring performance when the page is actually broken isn't very useful, so 
getting information about script errors that happened during the load is very 
important. This is most important when using WPT scripts, and when testing 
across browsers. 

On a similar note, when debugging these problems the console output is the 
primary (and only) way to log information from javascript out to the debugging 
tools. 

In short, it'll be great if WPT captured and displayed script errors and 
console output.
If necessary, this can be an optional feature, as sometimes capturing console 
output changes the application behavior.

When applying a software update the system is currently rebooted.  Change the 
default behavior to just restart urlblast.exe.

A reboot can be pretty expensive and in the case of EC2 the systems really 
don't reboot reliably consistently.

Chrome uses non-blocking writes. That means that the unencrypted write calls 
that we hook in hook_nspr.cc do not match up with the actual networking in 
hook_winsock.cc. In fact, I have seen cases where a GET request comes in the 
middle of an SSL handshake. A way to remedy this is to only collect the times 
from hook_winsock.cc. We may have to beef up our processing of SSL data. 
(Wireshark is able to recognize HTTP is going by, so it is possible).

http://www.google.com/codesearch#wRp03gCO51Q/epan/dissectors/packet-wtls.c&exact
_package=http://anonsvn.wireshark.org/wireshark/trunk&q=handshake&type=cs&l=127
http://anonsvn.wireshark.org/viewvc/trunk/epan/dissectors/packet-wtls.c?revision
=39276&view=markup

A nice benefit is that if the debug symbols fair to download, then you still 
see ssl traffic in the waterfall.

Get dummynet working on 64-bit windows

Firefox is on a 6 week release cycle now.  We need to find a way to have it 
auto-update so that we don't have to constantly manage the test systems 
manually.

Implement the basic http and video capture support for Firefox (3.6+)

DNS lookups are not currently being captured for Firefox 7

It would be nice, once going through the website and setting all of the 
parameters the way you need them, to have a means to automatically generate 
what the API URL would be for the test you are requesting.

Chrome and Firefox script commands will not follow an href if clicked because 
the click event itself is not enough to trigger it.  It works in IE because we 
use a different mechanism for triggering.

When resources are blocked by pagetest it would be helpful to be able to see 
what resources were blocked instead of them just disappearing from the 
waterfall.

The HAR export code assumed IE as the browser because that is all that was 
supported when it was created.  Now that Chrome and Firefox are supported it 
should report the actual browser used.

Create a Chrome extension for interactively testing WPT scripts (mostly just 
the navigation and form filling commands).

I'm not sure if I missed it, but it looks like <default> value is no longer set 
in locations list:
https://sites.google.com/a/webpagetest.org/docs/advanced-features/webpagetest-re
stful-apis#TOC-Location-information

This was quite helpful for ShowSlow which shows API endpoints dropdown based on 
this call - I do want to add an override in ShowSlow, but it would be great if 
WPT instance provided a reasonable default (right now it defaults to IE6 
because it's a first location in the list):
http://www.showslow.com/details/?url=http%3A%2F%2Fwww.google.com%2F#pagetest

Will be happy to test on my end when you're ready.

Sergey

In the repeat view case, add support for deleting items with an expires < a 
provided window to allow for a simulated repeat view that is sort of aged (not 
well but used by some customers)

Add support for capturing a version of the Chrome/Webkit dev tools timeline 
that can be loaded into a user's instance.

Fix spelling error

Eliminate the need to specify job directories in loactions.ini and create them 
automatically

Add support for intercepting decoded SSL traffic for Chrome