W3C Compliance Testing: Methodology

We are going to redesign a large web service which has around 25 unique page templates delivering thousands of pages. One of the major requirements is w3c compliance (html 4.01 transitional / WCAG1.0). T开发者_开发问答here are an infinite no. of validation errors as the code is shabby.

1) Since its an old application, starting from scratch is not an option as it will break the site.

2) Also, there are some parts (less important parts) of the site where there are no changes. Do you think we can validate this in the beginning.

3) Validating each and every page is going take too long and we may not have the time.

So my question is, What according to you is the best methodology to go about validation? How will you estimate for such a scenario.

Thanks in advance!

Vasanth

Decision making and prioritising

When aiming for validation across a large site it can be good to start off with a prioritisation task.

This will give you some time to find out which pages in the site are most often accessed, what do your real users use, and given that, which of those are used the most and would benefit from being made valid first.

Validation is a requirement to my mind, however being pragmatic in a business context is also of high value.

Do you know how much time you want to spend initially working on validating your website code? and given the time you've allocated, knowing which parts to start with will give you a real head start.

Estimating

If your team arn't used to writing valid code and have never undertaken a validation task, estimating is going to be hard for them. You can however estimate in terms of complexity points which will then give you more visibility and transparency to offer the business bods.

Ad-Hoc

Get used to correcting errors as you go. Have each developer on your team fix up 2 errors for each page they work on, each time they work on them.

Over time, you'll end up with a nicely validating website and combined with an initial revamp, this is a great way to keep pages valid, keep developers on top of validation and saves a lot of time in the future.

If there are many pages, a good start is to set up stuff so you can crawl your own site and validate all pages by a "single click operation". How to do it depends on your environment, but it is actually not that hard.

When you have a lot of errors, a good first step might be to create an even more "relaxed" schema than html 4.01 transitional, which allows some of your known errors to begin with, to get a grip on what you have to change. When the site validates with this extra sloppy schema, it is time to strict it up to html 4.01 transitional (and maybe all the way to 4.01 strict).

And the most important part: Try hard to not only have validating markup, but also relevant markup, so the heading structure of your site is sane, <em> is only used for emphasis and not for other reasons to have the text in italics, etc, tables are used for actual tabular data, etc.

Re estimation: I would estimate a day or three to get the "single click validation thingy" going, and before I got the output from that I would refuse to give any estimate at all about the whole fixing (unless I happened to have a really good grasp of what the actual problems are in the templates from the start).

3) Validating each and every page is going take too long and we may not have the time.

Then I don't think you're after a ‘methodology’. How about just installing the Firefox Validator add-on and having it enabled when you browse around your site to pick up errors as you go, starting in the most commonly-used pages?