It came as little surprise to hear that Autodesk were buying up Revit. At least, I had expected that they might like to do so, but I thought it unlikely such an offer would receive a warm response due to Revit being partly owned by Autodesk’s arch-rival, Dassault Systèmes of France, producers of the mighty CATIA and owners of SolidWorks and now also of the ACIS solid modelling kernel.
Autodesk’s interest in Revit relates to its need for a modern 3D object-based architectural system — indicated by the successful venture with Inventor in the mechanical field.
Following the advent of AutoCAD R13 but prior to Inventor’s introduction, the Autodesk vision was still based on specialised market products that added-onto the base AutoCAD using the ARX programming system, and its new facility for creating ‘Custom Objects’ to add to AutoCAD’s repertoire of drawing objects.
The main progeny of that vision have been ‘Autodesk Architectural Desktop’ (ADT) and ‘Mechanical Desktop’ (MDT). It became apparent in the time since these products emerged, that the ARX add-on idea would be neither permanent nor long-term as a satisfactory method for producing competitive specialist 3D design systems. Eventually, Inventor was released alongside MDT, presumably after a lengthy period of gestation in secrecy. Inventor has quickly usurped MDT, and not surprisingly in my opinion.
I think the ARX add-on concept has many drawbacks. Firstly, it depends on the line-oriented CAD system, AutoCAD, as a foundation, with its data handling system and DWG file format that was designed in the 80’s for a purely line-based system. Although it had a major revamp for R13, it still has to embody support for old DWG drawing files. I don’t know how much the big R13 data system rehash changed the old AutoCAD data structure within AutoCAD’s memory environment, but it certainly perpetuates the old data structures in its programming interfaces in LISP, ADS, ARX and DXF. The newer VBA (Microsoft Visual Basic for Applications) interface exhibits a different data structure format based on Microsoft’s COM standard for Windows, but just what the actual AutoCAD internal structures are, cannot be deduced.
Adding ‘Custom Objects’ by ARX and saving them within the DWG file leads to the now well-known ‘Proxy Objects’ when a DWG file produced from a system that creates Custom Objects, is opened in ordinary AutoCAD. Autodesk created a mechanism to alleviate this problem with their ‘Object Enabler’ technology, but it remains a confusing topic. The fact is, DWG files produced from ADT and MDT are in effect quite different from ‘ordinary’ DWG files. The whole aura of ‘Industry Standard’ associated with DWG is derived from the assumed interchangeability of DWG files, and Custom Object-based systems destroy that interchangeability.
The second problem with ARX add-ons is the fact that all the old 2D and line-based AutoCAD drawing tools are still there and usable alongside the new 3D and object-based tools. Although I know some ADT users do mix the two routinely, I am surprised they don’t run into problems more often. It certainly has potential for problems. Of course, it is the existence of the familiar old AutoCAD menus and tools, that makes ADT seem an easier pathway to 3D design for those used to AutoCAD. But users still have to learn a new approach to design if they are to use 3D methods. The presence of the old tools is not as much help as might be thought. Conversely, the AutoCAD environment is more likely to be seen as an impediment by those who are not used to AutoCAD. That was, I am sure, a major reason for developing Inventor.
Finally, the limitations of the old data management system from AutoCAD become apparent when you start to work with very large and complex 3D models, and with the ‘single building model’ design method, that is inevitable. ADT becomes exceedingly slow with large amounts of data, mainly I think because the AutoCAD data management system cannot shuffle data in and out of memory intelligently, so that only the data elements needed for rapid manipulation of the regions currently being worked upon, are held in memory at any one time. That ability is crucial for systems that must handle large assemblies, which is exactly what a building model is. Despite all the above, ADT has in fact done quite well, and I know of several offices that are using very effectively and are very happy with it. But I still think it is excessively complex and slow, and that a new system with a purpose-designed data system will be well received by AutoCAD architectural users interested in moving to the 3D method.
With the Inventor experiment proving the acceptability of a totally new and AutoCAD-free environment with a slick new command system, it seemed inevitable that Autodesk had to be looking to either developing a brand new Architectural 3D system or buying one. With the very long lead times of software development and its cost, the latter is always more attractive if a suitable product exists. Graphisoft’s ArchiCAD was the pioneer of the 3D single-model method for architecture (at least in an affordable manner) and little emerged to rival it for a long time. Eventually there was Revit, developed by people who had worked on the rock-solid Pro/Engineer manufacturing design system. After a shaky start, (at least from the non-US perspective ) it became a serious contender.
The other recent Autodesk acquisition, that of a ‘snapshot’ of the source code of the ACIS solid modelling kernel used in AutoCAD and Inventor, puzzles me somewhat in terms of its worthwhile-ness, but I have no doubt that Revit will be a very worthwhile addition to the stable.
