Link-testing methodology?

["Geoff Hart" <geoff-h -at- mtl -dot- feric -dot- ca>]

Chris Hamilton observed that <<... QA could require of us a 
method for assuring that not only are our links not broken, but that 
they point to the correct target. It seems to me that regardless of 
whether you automate this task or do it manually, you have to start 
by creating a list of each link on each page and its appropriate 
target.>>  

It should be easy enough to find the links; if your authoring 
software doesn't actually display a table of links, it should be 
possible to open the file in various word processors or text editors 
and use the search tools to locate all the codes that identify a link. 
But that's approaching the problem backwards: what you really 
want to do is create the equivalent of a design specification for 
each Web page or other online project that specifically lists what 
links you'll require. That list becomes your checklist when it comes 
time to do the QA thing: simply run down the checklist, confirm 
that each link is present, then follow the link to confirm that it 
actually goes to the right place.

If you've got a decent functional spec, that means you've carefully 
thought about what you need to link to, and the task of checking 
those links becomes relatively simple; if not, then don't dare 
complain about SMEs who don't stick to the specification! <g> 
Once you've got the spec defined, you can always go back and 
add links as you discover things that you missed... but you have to 
be rigorous about this (i.e., someone needs to update the spec and 
do a reality check to confirm that the new link is really necessary).

<<What's more, because of the fact that compiling such a list is 
likely to be a long, arduous task, someone should review it.>>

It's every bit as important to edit your links as it is to edit the text 
itself. But that's no more a task you can automate than you can 
automate the confirmation of index entries in printed 
documentation. Most authoring software (e.g., Dreamweaver for 
Web pages and RoboHelp for online help) now has the ability to 
quickly identify broken links, but someone human still has to make 
sure the link text makes sense (i.e., that it adequately describes 
the destination of the link) and that the link takes you to the right 
place. Until we develop software that can parse English as well as 
a human, no cyber-someone can make that kind of judgment. 
Fortunately, checking each link isn't rocket science, nor is it as 
difficult as it might seem. Simply take each page or file, one at a 
time, and follow each link as you encounter it. As soon as you've 
confirmed that the link is correct, hit the "back" button to return to 
where you took the detour. (Resist the temptation to follow a string 
of links; doing so is a sure way of getting so far from the original 
link you were testing that you forget where you started, and end up 
missing links.) If you do this for each file as soon as the file's been 
approved for "layout", then all your files will be checked within a 
short time after the overall project is over.

--Geoff Hart @8^{)}  geoff-h -at- mtl -dot- feric -dot- ca  (Pointe-Claire, Quebec)
"If you can't explain it to an 8-year-old, you don't understand it"--Albert Einstein