Notes on Link Checking / Site Reports


The following are notes on using the link checking / site reports provided on the redesign work-in-progress page.  Link checking and other maintenance should be completed by March 28th, to coincide with the completion of link research.  This maintenance is to fix what we already have online, not to implement any new content, or the new design.

 

1. In order to aid in the process of cleaning up pages, hunting down rogue files no longer in use, fixing broken links, and making pages load faster and work better, Linkbot v5.0 reports have been provided on the work-in-progress page.  These reports were performed on a directory basis, dependent on where a site is on the server’s directory structure.  For example, the report for the Department of Public Works and Environmental Services is in the ‘gov’ folder under ‘dpwes’, so the report is named ‘gov / dpwes’.

 

2. Within a frameset, you will initially see an overview of your site, with the number of specific problems that exist and their percentage relative to all the problems Linkbot found.  Note that internal links – links within our own site – and external links to pages outside Fairfax County’s – are separated. Slow pages are pages that, graphics included, are over 75kb and take “awhile” to load.  The “old pages” cutoff date has been set at one year.  Missing attributes would primarily consist of pages with images that lack alternative text, i.e. the ALT tag.  Orphaned files are files that, according to Linkbot, are no longer in use by the site it is examining and are therefore unnecessary and most likely should be removed.  Note that any file whose name has spaces in it (an example would be ‘big red balloon.jpg’) will show up no matter what – please ignore all orphan files with spaces in their names.  In the future, naming conventions for images and pages will be addressed by HTML / Web development guidelines.  Note that some areas of these reports, like the ‘Site Quality Rating,’ were not enabled and will show up as missing.  They are not necessary to complete these tasks, and were in many cases inaccurate.

 

3. Linkbot is not perfect.  Do not interpret these reports as 100% accurate and go deleting files left and right, or taking down pages it reports as slow, without first verifying the information yourself. In many cases, a javascript redirect can confuse Linkbot, as can the aforementioned spaces in filenames.  Also be aware that in checking external links, Linkbot only waits so long for the server to respond.  There may be external links on your site to pages outside the county that load very slowly that it will consider ‘broken’.  Use your own judgment in assessing whether such links should still be up.

 

4. Begin with fixing broken links, with a priority on addressing internal links first, then external.  These actions can be taken immediately, and do not (and should not) have to be held off until the redesign goes up.  In this process, you won’t be modifying the pages to fit the new look, but simply fixing what’s already there.  In instances where a link is broken and it is external, or to another agency’s material on our own site, try to find a replacement link, consulting whoever originally put your content together as well as the party being linked to.

 

5. Addressing slow pages is a bit trickier than fixing links, as 9 times out of 10 the problem is not with the page itself but the graphics on it.  From observation, we have noticed that the largest unnecessary bandwidth on our site is from pages that scale images using html.  In many instances, images that are 100K or more are being scaled down to thumbnail size using the WIDTH and HEIGHT tags.  This is NOT the proper way to scale images.  Images should be scaled to the desired width and height in an image editing application like Photoshop, Paint Shop, or any number of shareware and even freeware programs.  Some pages will be over the 75K limit and can remain so if everything has been optimized as much as possible, but evaluate every link that Linkbot reports as slow and see what can be done.  If a page has to remain over the limit, try to warn visitors of the load time before they visit that page, with the page’s total size in Kilobytes next to the link, in parentheses (i.e. (120K)).  A contract is in place to provide vendor support to those who need it in optimizing / resizing images.

 

6. The w3c’s Web Content Accessibility Guidelines clearly state the need for ALT tags on all images in order to allow users with disabilities or text-only browsers to navigate your pages.  Linkbot reports pages that lack such tags under ‘Missing Attributes’.  Correcting every single missing attribute error is probably not feasible before our June deadline; however, on pages that are absolutely critical, and especially on images that are necessary to understanding a page’s content, ensure that ALT tags that describe the image are present.  These ALT tags should be more than simply the filename and size, but should describe (basically) what the image is and any textual information it conveys as well.  For image maps, make sure that every hotspot has its own ALT tag as well as a separate ALT tag for the entire map itself.  For images that are not important and are used solely for design purposes (fancy horizontal bars, for example) use a null ALT tag – ALT=“”

 

7. Orphan files are the biggest reason our search engine returns results that are confusing and useless to visitors of our site.  Orphan files are files that are in your directory that, according to Linkbot, are no longer being linked to and should be removed.  In addition to unnecessarily taking up space on our web server, any of these files that are searchable (html, doc, txt, and pdf) will be searched and indexed by our search engine.  In some cases these are older versions of pages that were never deleted when new versions went up, but were just renamed and forgotten.  In other cases these are areas that were being used to test new ideas but never got cleaned up after the fact.  In any case, wherever possible, these files should be copied over to your local machine and deleted off of the live server.  Do not delete anything until you are absolutely sure it is no longer in use.  Linkbot picks up some pages that aren’t really orphans due to technicalities – pages that have spaces in their filenames, or are default.htm or homepage.htm documents that are never referred to by name (for example, all references to

http://www.co.fairfax.va.us/library/

 

that leave off the ‘homepage.htm’), and pages that are only referred to when an anchor is on the end (for example http://www.co.fairfax.va.us/gov/dpwes/construction/codeload.htm#codes)

 

will show up as orphans but aren’t.  Priority should be given to removing searchable files before images, as orphan image files at least do not affect the search.  While addressing all the orphan files on your site might be too time-consuming given the other tasks involved with the redesign, do your best to remove what you can, when you are sure it is appropriate to do so.

 

8. If you have any additional questions or concerns, or need a report run on a specific area of the site that is not yet available, please email David Lloyd.  Reports will be run again on March 14th for comparison.