A public forum for discussing the design of software, from the user interface to the code architecture. Now closed.
We are about to commit a conversion project in which we are required to convert a old VB 6.0 application to Microsoft .net 2.0 platform. The old legacy applications revolve around old COM + as a middlewayer. The user interface layer communicates with the COM+ components using XML through RPC (COM+ components are hosted on a application server).
When devising migration strategy we encounter number of forums and articles advising to re-write the code as the whole conversion process is buggy and requires lots of efforts to fill the errors occurring due to failed conversion.
Apart from this when we analyzed the business case we realized that converting the whole code would result in a .net application based on outdated architecture which would mean huge change cycles and same old extendability issue re-occurence.
We would appreciate an expert advice of community on this before drawing any conclusion
Your options range on a spectrum between these two points:
On one end, you do the conversion as conservatively as possible, changing only language, and leaving architecture and design as-is. Benefits: you have a plan to help protect you from scope creep. Drawbacks: you're likely to end up with all the design and architecture flaws you've got in your current system, plus you may not be taking advantage of framework features in .Net.
On the other end, you do a complete rewrite, treating the current system as a prototype. Benefits: best possible opportunity for "if I were to do it all over again, I'd do it differently". Drawbacks: it's real easy for this to turn into a gaping black hole from which you may never return.
Either end of this spectrum probably doesn't make sense for you. The "right" middle ground will likely depend on business factors such as budget and timeline, and on the state of your current system (a sound, modular design will lend itself to a more literal conversion, which may be less risky).
I'd look for places where you can draw boxes around functionality and convert these components either first or last, so you can isolate risk into smaller chunks of work.
Wednesday, October 03, 2007
I agree with the previous poster. Almost anybody on the post who do VB6 to .NET conversions face the same problem - a complete rewrite. If you can, do it but do you have the resources for it now and do you want to do a complete rewrite now ?
Thursday, October 25, 2007
This topic is archived. No further replies will be accepted.Other recent topics
Powered by FogBugz