Tools We Use: VSE Link Tester
Gone are the days when you could easily build and maintain a Web site using nothing more than SimpleText, NCSA Mosaic, and a rough mental image of how pages linked together. On today’s Web, it’s not uncommon to find yourself lord of a sprawling Web metropolis that sprang from seemingly humble beginnings. Now, with thousands of links referencing both internal and external pages, the scope of maintaining those links has progressed beyond what one person can do.
Fortunately, this "beyond mere mortal" stage is often when good utilities emerge, hints of the promise that computers could make our work lives less repetitive and more rewarding. When I need to make sure a client’s site is navigationally sound, I turn to VSE Link Tester 2.5, an application that not only checks links but makes it easy to track down and fix the errant code.
<http://vse-online.com/link-tester/>
Break Test — The concept behind Link Tester is simple: access a Web page like any other browser, then follow every link to make sure it returns a valid object such as a Web page, image, or downloadable file. Link Tester can also test JavaScript links and events like those used for image rollovers. Unlike a Web browser, though, Link Tester opens several simultaneous connections to speed up the workflow. When it finds a broken link, Link Tester adds the affected file name and the location of the bad link to its report. You can tell Link Tester to check just the links on the page you specify or to follow links recursively, which makes Link Tester check the links on other pages at your site that are referenced from the main page. By default, the program follows links to other sites as well, so you can remove or correct dead links.
Of course, you could do all this manually. For hours on end. Clicking until your fingers go numb and your eyes turn to jelly. But I prefer to run Link Tester and go enjoy a cup of coffee.
When Link Tester has followed all the links, it builds a HTML-formatted report detailing the links that were checked, which were broken, and the reason why they didn’t work. The program even includes an Error Explanation window that lists and explains the most common problems encountered.
Strengths — Link Tester understands how people use the program, and throws in just enough extra functionality to appeal to a broad range of users. Every site you scan is stored in a master list in the main window, so it’s simple to go back and re-run previous tests. You can also scan local files offline, specify the filename used when the URL ends in a slash (such as index.html or default.html), and be conscious of case-sensitive URLs on some systems. A helpful new feature is the capability to create filters to ignore addresses; for example, it can skip past URLs that are stored on a different machine when you’re testing offline.
When testing remote links, Link Tester includes a modicum of control over how it interacts with Web servers by offering a Server Load setting spanning five steps between Very High and Very Low. Although the interface is ambiguous, in practice Link Tester opens fewer connections to remote servers at lower settings.
Weak and Missing Links — From the point of view of a Web server, though, Link Tester’s method of opening multiple simultaneous connections can be problematic. Even at its lowest Server Load setting, Link Tester requests files much faster than a real user; at higher settings some Web servers will interpret Link Tester’s accesses as a denial-of-service attack. If you send Link Tester recursively into a large or infinite URL space (like the TidBITS article database), it will happily pummel the remote server for hours, or even days; further Link Tester doesn’t obey robots exclusion protocols or META tags, so even if webmasters mark those areas as off-limits to automated programs, Link Tester won’t notice. For best results, use only the lowest Server Load setting when checking links to any sites other than your own.
<https://tidbits.com/getbits.acgi?tbart=00946>
Another potential annoyance is the way Link Tester creates its reports. Each test is saved to an HTML file within a folder named using the URL and a number (such as "www.jeffcarlson.com 001"). Each report folder contains an images folder with a handful of icons used in the report. So, whenever you create a new report, Link Tester clutters your drive with a new set of identical images. It should be just as easy to store these images in one place and reference them in the reports.
I’d also love to see Link Tester support scheduling tests for automatic execution. This is just the type of tool I’d love to park on my PowerBook 5300cs (now acting as a Retrospect backup server) and have run in the middle of the night.
Thinking about Linking — I like Link Tester because it’s straightforward and powerful: typically, after a few minutes I can track down an errant URL or help unravel why something isn’t displaying.
Link Tester 2.5 is available in two editions. The standard version, which costs $20, will search one URL, following an unlimited number of links from up to 20 pages on your site. The Business version, at $80, can test an unlimited number of links and pages; an Academic version with the same functionality is available for $40. The unregistered software lets you enter one URL, and provides a limited error report. The software is a 1 MB download. Link Tester requires a 68K or PowerPC-based Mac running System 7.5 or later.
<http://vse-online.com/link-tester/download.html>
[11-Sep-000 — When this items was originally published, in many places it incorrectly said the product was named VSE Link Checker, rather than VSE Link Tester. We’ve amended the text here, and published a correction in TidBITS 547.]