Maybe a glitch of the website - it misbehaves at times. Anyhow, I could read most of it in my e-mail. Here it is:
I'm considering Scilab/Octave/Sage (ie clones of Matlab) for their total scriptability. This is the area where Mathcad stopped its development. Suppose you make a generic Mathcad file with calculations of a standard product, but the product has few versions. I don't know, let's say a pressure vessel can have 2 or 3 nozzles. You make 3 chapters, each with all calculations for the nozzle, and simply hide the 3rd section when it's not needed. Then you realise that there are hundreds of these options, some calculations are only applicable when other conditions are on or off- it stops being manually controllable (and the TOC is a nightmare). You need automatic scripting to do all this logic- on the level higher than the document itself. Mathcad is not an option- you can't control the document this way.
Another example is naming variables. Following design codes (particularly when there's more than one code in a job), you always get few different variables with the same name. No problem: you add some prefix, postfix, ie you change the name a bit. Then a stubborn inspector starts winging: it's not what the code book says, your formula looks different. Sure, you can't use any aliases, or local variable names. Your P is the same P from page 1 to the end, and if you redefine it- it's lost.
There's more fundamental problems which can't be solved with Mathcad, at will never be with the current pace of development. These are not mathematical problems, just the way your concept is presented on paper. Definitely, not the highest priority for PTC.
I remember my first comments to Andrey after I saw an early version of his SMath- similar problems of presentability, clarity, etc. He took it seriously and addressed it ages ago...
