Fw: Web site QA and testing

Here is a response to my query about testing of localized web sites,
for those of you who are interested.

----- Original Message -----
From: Mike Brenner <mikeb@mitre.org>
To: Suzanne Topping <stopping@rochester.rr.com>
Sent: Monday, September 27, 1999 2:18 PM
Subject: Re: Web site QA and testing


We communicate with people who use various languages including
Japanese, Hebrew, English, German, Korean, Arabic, Hindi, Russian,
Cantonese, Mandarin, Italian, Danish, Norwegian, Swedish, Finnish,
Polish, Bulgarian, Romanian, Czech, Serbian, Croation, Albanian,
Greek, Dutch, Vietnamese, Thai, Urdu, Farsi, Kurdish, Turkish,
Turkik, Ukranian, Indonesian, Cambodian, Estonian, Lithanian,
Latvian, Gaelic, Portuguese, Afrikaans, Zulu, Armenian, Catalan,
etc., etc. We have the technology to display some of these languages
on web sites, but others give problems. All of these languages give
problems when we wish to have a given web site in a large variety of
languages.

When will all web browsers come with all Unicode fonts and
tranliterations built in?

I use the following guidelines for testing web sites.

A. How many clicks (keyboard or mouse) does it take to bring up a
web page in one window in English and in the other window in the
customer's language and alphabet? (Since we do not have the
technology to do this for some languages, the answer is sometimes an
infinite number of keystrokes is not sufficient; instead, we must
hire a manual translator and then scan the translation on-line in
graphics.) GOAL: 2 clicks or less.

B. How many clicks (keyboard or mouse) does it take to install the
browser modifications for each language? (This includes the clicks
to install any required fonts. If we cannot do it in a particular
language the answer is infinite.) GOAL: 12 clicks or less.

C. Has the correctness of translations been tested by going from
language A to language B and then back to A again, so that web pages
without intentional vagueness (like political rhetoric or sales
presentations) are verified to a basic level? GOAL: Yes.

D. How many clicks (keyboard or mouse) does it take to use the naïve
language-translation software to get an approximate translation of a
web page from language A to language B and then back again to
language A? (In many combinations, A and B, we do not have automatic
translation software of any kind and the answer is infinite). GOAL:
2 clicks (one click each way).

E. Are all links in the web page accessible when hosted on a Solaris
Server? (Frames, Broken links, Basicscript, Microsoft-Only,
Macinstosh-Only, and other non-transportable things are measured
here.) GOAL: YEs.

F. Are all links in the web page accessible when hosted on a
Microsoft? (Broken links, Sun-Only, Microsoft-Only, Macinstosh-Only,
and other non-transportable things are measured here.) GOAL: Yes.

G. Does the email of the person responsible for updating each web
page appear at the bottom of the page with a hyperlink to their
e-mail address? GOAL: Yes.

H. How many seconds does it take to finish displaying all the text
on the page? (If there are any advertisements or pictures or if the
text had to be displayed as a gif or jpg, or if Java programs take
more than a second, etc.) GOAL: 1 second or less.

I. How many of the main keywords of a page are hyperlinked to
related pages inside or outside of the web site? Goal: 12.

J. Does all text wrap correctly? (This fails for JPG and GIF
transliterations) GOAL: Yes.

K. Are all the main keywords and all the meta-data correctly
annotated in XML? GOAL: Yes.

L. Are all the table headers correctly annotated in XML? GOAL: Yes.

M. Are all of the pages accessible to anyone who has access to the
search engines we permitted to web-crawl the pages? GOAL: Yes.

N. Does each web page have a unique name or number to refer it by
when reporting spelling and factual errors, or suggesting
enhancements and new links? GOAL: Yes.

O. After reading the web page, did we get all the information
promised by the title of the page? (A common example is an
e-commerce site that has a lot of information about the company, but
forgot to put their catalog and price list on-line, so they make no
sales from their web pages.) GOAL: Yes.

P. Is there a series of links originating on each web page that
leads you to an address, phone number, fax number, and e-mail of
someone who can sell you the stuff described on the page. (A common
example is a government site that describes the service they
provide, but other government agencies who use those services cannot
determine a point of contact from reading the web page.) GOAL: Yes.

Q. Are all the correct language translations linked to each page?
(For example, Canadians do not forget to make English and French
links on their pages. But US pages selling products to other
countries often do not have links to translations into those
countries' languages.) GOAL: Yes.

R. Are all non-ascii languages linked to ASCII transliteration
pages, until the time when common browsers like Netscape and
International Explorer can correctly display all the characters?
GOAL: Yes.

Mike Brenner


Suzanne Topping wrote:
> I am researching requirements for QA and testing processes for
> localized web sites. I would appreciate any help you can give me
> if you know of any existing resources, or if you've developed your
> own guidelines that I could review.

Received on Tuesday, 28 September 1999 10:14:08 UTC