Top Banner
Evaluation and testing 1 WUCM1
34

Evaluation and testing

Dec 30, 2015

Download

Documents

cassidy-bolton

Evaluation and testing. Testing large websites. Testing should be against agreed requirements For large sites with dynamic pages (Java, ASP, Perl, etc.): Use a development environment such as NetBeans, DreamWeaver or Visual Studio - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Evaluation and testing

Evaluation and testing

1WUCM1

Page 2: Evaluation and testing

Testing large websites

• Testing should be against agreed requirements

• For large sites with dynamic pages (Java, ASP, Perl, etc.):– Use a development environment such as

NetBeans, DreamWeaver or Visual Studio– Offers debugging services, break points, traces,

variable contents, spying, etc. – Permits remote debugging

WUCM1 2

Page 3: Evaluation and testing

Testing small websites• Start with fully specified checklist

of expected functionality• Greenberg (1999) gives a good

example.

[] 1. Main page loads without errors

[ ] 2. Banner frame loads banner[ ] 3. Hover functionality over each of

the banner links[ ] 4. “About” is a link to about.htm[ ] 5. “Feedback” is a link to

feedback.htm[ ] 6. “Advertise” is a link to ad.htm[ ] 7. “Awards” is a link to

awards.htm[ ] 8. “Rings” is a link to rings.htmetc.

WUCM1 3

Page 4: Evaluation and testing

Browser compatibility tests• Important to test your website using a variety of

browsers– cannot assume it will look/operate the same in all

• Use a range of different browsers and browser versions• Browsers you do not have, e.g. for a Mac? (

www.danvine.com/icapture) shows the ‘look’• Some tools permit you to exercise a range of browsers

– e.g. see the compatibility viewer http://www.delorie.com/web/wpbcv.html

• How do you record your test results? • Very least is a tick in the box, as above

WUCM1 4

Page 5: Evaluation and testing

Web server tests

• Most tests presume the error is in the HTML, ASP, Perl, JavaScript etc.

• Possible that there are faults in the web server configuration

• How to modify site test plan to investigate the server environment?

WUCM1 5

Page 6: Evaluation and testing

Link tracking and checking 1

• Tests should involve the link integrity of the web structure

• Web development systems (e.g. DreamWeaver) usually include a Link Checking option

• Link checking:– Validates your structure– Highlights broken links– Identifies orphan files – Lists external links

WUCM1 6

Page 7: Evaluation and testing

DreamWeaver example

WUCM1 7

Page 8: Evaluation and testing

Link tracking and checking 2

• Employ a 'web weeder' to check for:– Out of date links – Out of date documents

• Check external links:– Tedious to do manually– Automated tools make this chore easier– See the Web Design Group's links page

http://www.htmlhelp.com/links/validators.htm

WUCM1 8

Page 9: Evaluation and testing

Link checking tools 1

• Options for installation:– On your web server– On the service provider’s server– On your development machine

• Examples of external services:– The fee based WebAlarm system

• http://www.linkalarm.com/– NetMechanic

• http://www.netmechanic.com/

WUCM1 9

Page 10: Evaluation and testing

Examples

WUCM1 10

Page 11: Evaluation and testing

Link checking tools 2

• Example – download link checker:– Xenu’s Link Sleuth

• http://home.snafu.de/tilman/xenulink.html• Someone needs to react to the Link Check

report:– Updating links if the service has moved– Rewriting the pages if the target has been retired– Advising users of problems in the past, so at least

your users are forewarned

WUCM1 11

Page 12: Evaluation and testing

Link checking tools 3

• Users who tried to use your broken links– Use a tool to identify them – 404 errors– Contact them (if identifiable)

• might be good PR?

– WatchFire used to offer a range of white papers• http://www.watchfire.com/

– Check on Google

WUCM1 12

Page 13: Evaluation and testing

Use of log files

• Some common log file formats:– Common– Referrer– Agent

• User defined possible:– Would need you to produce your own analysis

software– Seldom worth the trouble

WUCM1 13

Page 14: Evaluation and testing

Log files 1

• Text in the format string not a place holder; is just sent as plain text

• Placeholders are replaced with their data– LogFormat "Host=%h URL=%U Server=%V Port=%p”

• Log files can be useful to look through, but you will need help

WUCM1 14

Page 15: Evaluation and testing

Log files 2

• Simplest use is to just filter and display the most recent entries – – suitable for a human to monitor

• Benefits of visual inspection:– Check that pages deliverable (no 404 errors)– Check that security is working properly– Check that CGI is working properly– …

WUCM1 15

Page 16: Evaluation and testing

Session logs 1

• Logs that track a single user session– What pages were viewed?– What order were the pages viewed in?– Was a purchase made (or some other goal

reached)?– Was enquiry abandoned?

• Not easy to achieve as web inherently “stateless” – each page is a new transaction!

WUCM1 16

Page 17: Evaluation and testing

Session logs 2

• Techniques – see Nobles (2001):– Cookies – using cookies to capture and log session

data– Database-driven pages – using the code to capture

session data– Parsing server logs [next slide]– Web bugs – hidden <img> tagged files that trigger

unseen activity at the server– often 1 pixel square

– Non-covert <img> tracking – the more polite version of web bugs

WUCM1 17

Page 18: Evaluation and testing

Log file analysis tools

• Many tools available – try a Google search• Analog (free from http://www.analog.cx) is

typical – see Wainwright or Nobles• Analog system is configured with a .cfg file

much like Apache’s• Some examples from the website

WUCM1 18

Page 19: Evaluation and testing

Analog examples 1

WUCM1 19

Page 20: Evaluation and testing

Analog examples 2

WUCM1 20

Page 21: Evaluation and testing

Analog examples 3

WUCM1 21

Page 22: Evaluation and testing

Analog examples 4

WUCM1 22

Page 23: Evaluation and testing

Webalizer example 1

WUCM1 23

Page 24: Evaluation and testing

Webalizer example 2

WUCM1 24

Page 25: Evaluation and testing

Uses of log file analysis

• Qualitative checks:– identify (and then eliminate) any 404 file not

found log entries• Nobles (2001) lists a number of "quantitative"

uses:– How many page views is your site getting per day? – How many user sessions is your site getting per

day? – How long are users staying on your pages?

WUCM1 25

Page 26: Evaluation and testing

More uses

– Which pages of your site are requested the most? – Which pages are requested the least? – What are the top ENTRY pages for your site? – What are your top EXIT pages? – Which pages are being viewed by themselves,

where visitors are not even clicking to go to another page?

WUCM1 26

Page 27: Evaluation and testing

Even more uses

– Are visitors from other countries visiting your site? – Which sites are sending you the most visitors? – What are your top-referring search engines? – Which keywords are searchers using to find your

site? – How many of your visitors are using older

browsers? – Which search engine spiders have visited your site

recently?

WUCM1 27

Page 28: Evaluation and testing

Web usage statistics

• Web statistics suffer from a number of problems:– Caching and proxy servers hide activity– Some users hide behind “anonymising” services– Most web pages built from very many small files –

how to count?

• Distinction between pages and hits

WUCM1 28

Page 29: Evaluation and testing

Load testing

• Need a feel for the behaviour of your web server under different levels of load

• A baseline measurement is vital for:– planning for growth– identifying and dealing with DoS attacks

• Tools include:– Mercury Interactive’s Loadrunner – RSW Software’s E-Load – Segue Software’s Silkperformer

WUCM1 29

Page 30: Evaluation and testing

Evaluation

• Checking to see if the website can be used effectively by its intended audience

• Evaluation in most cases will involve a human in the loop

• For a good technical set of examples see Spool, (1999)

• Not the main focus for WUCM1

WUCM1 30

Page 31: Evaluation and testing

Search engines/directories

• How will search engines and directory services interact with your site?

• Nobles (2001) considers the matter from a number of perspectives, e.g.:– Key words and interaction with spiders or robots– Optimising your site to facilitate search engine

spiders• Deprecated features of HTML

– Web directories and how to get included

WUCM1 31

Page 32: Evaluation and testing

Validation services 1

• Validators can for example check for a number of different features, including: – How usable are your pages to people with

disabilities– Conformance to the HTML specification (World

Wide Web Consortium - W3C.)

• See The Web Design Group list of links to different validation services– http://www.htmlhelp.com/links/validators.htm

WUCM1 32

Page 33: Evaluation and testing

Validation services 2

• Generally two ways of using a validation service: – Submit your page to the validation service website

and receive back a report page– Mount a copy of the validation software on your

development system, and run your pages through this system

• Like virus checkers, web validators need regular updates

WUCM1 33

Page 34: Evaluation and testing

Validation services 3

• Two important validation services are:– The W3C HTML conformance validation

• http://validator.w3.org/

– The accessibility validation services• http://www.cast.org/bobby/

– If you pass you get a logo, e.g.

• Review access to 1000 commercial websites– http://www.drc-gb.org/publicationsandreports/2.pdf

WUCM1 34