FireFox 0.9x

I tend to keep going back and trying out the Mozilla browser offerings simply to check my html against another browser when something odd is happening. I have never been partial to it or it's variants but I have to say the current FireFox browserversion is simply very nicely done. It even provides a handful of useful extensions that make html development a bit easier. It really is worth having a look if you haven't already seen or used it.

I know it won't be replacing my IE any time soon if at all.All the clients of my clients have standardized on IE and quite simply I still find it very programmable and a very useful tool (security issues aside). I am looking forward to the results of the new upstart in development attention the IE browser is getting from Microsoft.

I certainly don't long for the days of the browser wars with major updates all the time but some good healthly competition will certainly improve the baseline itself.

Just a quick aside; FireFox was orginally called FireBird but that conflicted with the Open Source InterBase software project. It is nice to see them change the name since the InterBase group had already been there.

JScript and Equality

It is just great having someone like Eric blogging on the details of the ActiveX engines. In particular this time he covers of some of the obscurity (ok, that is what I would call it) of equality comparison operators in JScript.

Read more on it here.

DOCTYPE, If your memory is as bad as mine

I know I personally go hunting over my hard drive every time I want to use a DOCTYPE declaration in an xml/html/xhtml document. If you have the same problem then you should bookmark this page.

A good resouce for JScript/JavaScript

If you happen to make use of JScript/ECMAScript/JavaScript then here is a good FAQ. It has some great tips and generally provides a good bit of getting started information.

Some good news for OpenSSL

OpenSSL, a great library for adding SSL and TLS support to your applications have received a good boost from the US DoD. The DoD went through the process of providing a series of validations for FIPS 140-2. In case you were wondering, that happens to be important to us Canadians as well.

Read more about it here.

Short-circuit evaluation got me

Virtually everything we do is server based on XML services. The majority is custom XML messaging versus using SOAP (in many cases we can easily outperform the standard SOAP stack by a factor of 10-35%). We are in the process of upgrading one of our major services and the existing test cases are sitting in individual HTML pages and using the Microsoft XML components to submit the incoming data and the user visually compares the results to the displayed expected outcomes (not very automated but it derived from the development itself). We are performing a major upgrade to this technology and one of the areas that we wanted to improve on was in the automated testing area. We have over 130 test cases for this particular server and it really isn't enough but in order to engage the user we need something that is friendlier than html and xml with a smattering of JScript. Perhaps stupidly, I asked for this particular task since it is something I wanted to further on a more general basis for the rest of our services.

Hmmm, lots of background for a lesson about even old dogs get caught by simple things. I have been using Pascal/Delphi since tp5.

Anyway, I have put together a specialized editor to allow the entry of the test case materials broken down by category, procedures, and individual test cases. This data is all managed using XML and client data sets (which are perhaps one of my most loved Delphi features). I use that data to generate a test file in a format that the primary tool I created, the HTTP/XML test runner, takes as input. It is with the runner tool that I had a very stupid mistake (no one has seen it, so at least some face has been saved ;)).

Effectively, I need to loop through the structured data (categories, procedures, and cases) and I do it using a simple while loop while maintaining the pass/failure status. The root problem itself is that this is a command line tool and I wanted the exit code to represent whether failures occurred to support automated build/test processes. As such, the processing loops are contained within functions which return false in the event that any inner tests had failed. This was being maintained by the deceptively simple "Result := Result and ProcessCase();". Sure, that works, looks like it works, but it doesn't. While testing I found that the process would give up the ghost as soon as the first failure would occur but continue to appear to work, but no more test cases would be run (hmmm).

Hey, wait, my side effect, the call to ProcessCase(), was not happening. Well, a big DUH! Back to the grind and thank goodness for failing test cases at the right time.