1. July 30, 2001:
Function(s) affected: vsip_ccfft
Page(s) affected: 515

The example for vsip_ccfft uses cblockbind as follows:

block = vsip_cblockbind_d(data, NULL, 2*N, NOHINT); 

block = vsip_cblockbind_d(data, NULL, N, NOHINT); 

It appears that all the symbol font characters in many equations throughout the pdf version of the API document are having their right edges cut off. An example is the Vector Ramp function on page 463 of the VSIPL API 1.01. In the "Functionality" field, the algorithm displays as Rk <- a - k * b, whereas it should be Rk <- a + k * b. In this case, the "+" symbol is being truncated to only its leftmost portion, so that it displays as if it were a "-" symbol. This appears to be a flaw arising during the conversion of the source document (which is in Microsoft Word format) into pdf form. We are trying to develop a procedure for conversion that will work more reliably with the equations but have so far been unsuccessful. The version 1.01 document currently on the web site has been updated by hand-editing of the pdf file, however, we cannot guarantee that all such rendering errors have beencorrected. We hope that the version 1.1 API document, expected to be available in the Fall, will have these errors corrected systematically by an improved conversion process. 

Errata first reported July 10, 2001 by Jason Hills, jason.hills@lmco.com. 

The functionality paragraph describes crfft2dop as a real to complex (inverse) 2D FFT. This should be stated as a complex to real (inverse) 2D FFT. 

Errata first reported July 12, 2001 by Brandon May, bbm@andrew.cmu.edu. 

In the section entitled "Data Memory Allocation", the last sentence of the 2nd to last paragraph currently reads: 

"The process of moving data from the logical VSIPL data space to the logical user data space is called admission; the reverse process is called release." 

"The process of moving data from the logical user data space to the logical VSIPL data space is called admission; the reverse process is called release."

In the section entitled "SUMMARY OF VSIPL TYPES", the third bullet from the bottom currently reads:

"_ul (, unsigned long int)" 

"_ul (unsigned long int)"

In the section entitled "Block Object", the comment associated with the assignment VSIP_CMPLX_NONE = 2 currently reads:

"No, preferred storage layout"

"No preferred storage layout"