It's not really true to state as this article does that the syntax was generated from the model 'as an after thought'. In fact a large number of the semantic constraints were enforced through the syntax though the mapping in 3 0 0 was fairly poor. The additions made in 4 0 0 had a much more direct mapping from the information model itself. The article also fails to note that a semantic validator was developed that validated the files generated by the various different tools against the specification of the information model was developed for 3 0 0 and later expanded to cover 4 0 0. This tool was developed by human understanding of the EXPRESS model rather than auto-generated from it.
It is not the case that 'most constraints ended up just being described as comments' there are some defined as such, but a very small proportion of the total. The statement that 'Most of the others became elaborate formal descriptions which most readers will never be able to decipher, and therefore may not stand up to automated debugging/compiling' is more defensible. Certainly I found that there were a large number of constraints expressed in conceptually clear but computationally inefficient forms, however I don't remember this reflecting the majority. I leave to someone else to decide if 'most readers will never be able to decipher' them. I'll note here that I didn't work on the modelling itself, merely on the validator.
Anon: This page could mention more about converters. eg: EDIF2NGD converts an industry-standard EDIF netlist to an NGO file--a Xilinx-specific format. http://toolbox.xilinx.com/docsan/xilinx4/data/docs/dev/apxb2.html
Lothartklein 20:22, 21 June 2007 (UTC): I'm as well not happy with the section "Evolution". As far as I can remember the stuff on "after thought" is only half true (or wrong). The initial development after 2 0 0 was first syntax based, but I though that already for the intermediate version 2 9 0 a kind of an Express based information model was available (I could be wrong here - I no longer have these old papers). It must also be noted that till its publication in late 1994 Express changes several times significantly. So at the time of releasing EDIF 3 0 0 Express was not fully ready. And only some 10 years later tools became available to really validate data sets for complex rules. I would vote for re-writing clause "Evolution" completely and separate facts from opinions (e.g. "... is a grand work, and a bedrock for discussions ..."). Maybe someone want to start.
History and future
It seems to me that EDIF is rarely used nowadays. Why? Has´ve other standards, languages or tools taken over, and if so, what are they? Or has the need for EDIF simply disappeared? If so, why? --HelgeStenstrom (talk) 11:48, 5 November 2009 (UTC)