ACCEPTED SOLUTION
Accepted Solutions
I can confirm the error in Mathcad 14, M045.
Its the vec2str(W) statement in the READCSVI routine which throws the error because it obviously can't convert a vector containing a zero into a string.
@Luc
Haven't digged into the logic of your program and can't tell why you set W to a 1x1 matrix containing a zero from time to time.
How does vec2str handles a zero byte in MC11? Maybe as a workaround we could write a new vec2str in MC11 which simulates that behaviour.^
!!! EDIT:
Solution found!
Simply replace the two occurrences of W <-- (0) by W <-- 0 and it seems to work OK. At least it works in MC15
I can't give it a try with P4, but I guess the problem there is the very same.
Below you see the difference between vec2str(0) and vec2str of a 1x1 matrix containing 0. The first returns an empty string (which is needed) but the second throws an error. Guess this is different in MC11
Hi The Cfg now works but Dat file says : "can't open the data file"Werner,
I remembered once having had a bad experience with supplying an integer to the vec2str() function.
(The error message says that the argument must be a vector)
But I've checked vec2str() can live with a zero, also in Mathcad 11:
So I could change single element vector containing a zero into a simple zero.
A message to the OP:
In Prime you might not need the READCSV function, because Prime has READCSV, which does the same in this application.
Luc
I also corrected an error, which means that the data file is now read including the very last sample
(I had to omit the last sample because the function was apparently reading past the end of the data file).
Attached are corrected Mathcad 11 and Prime 4 files.
I don't think that the corrected Prime 4 file will solve the problem you show above.
Luc
Ah, I see from your screenshot that this
is the difference to later Mathcad version which throw an error (maybe we could call that a bug).
Still not sure why the OP now has problems reading the *.dat file.
Only explanation I can thinks of is, that he forgot to copy it to the same locations as the cfg and MC file. Also possible that the file is open and in use by another program - not sure if MC would choke on hat.
I'm afraid I can't solve the problem that Prime has with my (translated) function. I've started my 30 day evaluation period in order to be able to run programs in Prime 4. It appears that Prime has a big problem with one of the statements in the READCSV program.
I've checked all tricks that the statement uses, and verified that they work in (other) Prime test programs that I made. So that is not the problem. The values of r, c and W just before the statement are as expected, and Prime does allow such assignments to R, effectively creating the matrix. So I cannot see why this should be a problem.
I think that simply Prime 4 is the problem: "system error"...
Anyway, it works in Mathcad 15, I suggest you use that to analyse the Comtrade data.
Success!
Luc
I agree that the problem seems to be buggy Prime.
Same error here with Prime 3.0.
Crazy enough that the following workaround(?) solves the problem:
At least it works in Prime 3.0:
Yes, same error as in Mathcad 15 as I noted above.
OK, maybe its not to be seen as bug as a string containing the NUL character is not the same as a string containing no character at all, but the error message is misleading and it would be quite convenient if vec2str would accept a scalar as argument.
But this vec2str behavior does not explain why my workaround with the if statement is necessary. It should work as it was written by you because vec2str(0) returns the empty string anyway. First idea was that it isn't en empty string but rather a string containing an invisible NUL character, but Mathcad told me otherwise
So I believe that the strange error message we get in Prime is the result of a a Prime bug.
Anyway, our "Mathcad User" should now be able to use your program in Prime, too. He seems to have another Problem (in MC15) when using a different input file.
