Version 4.0 | |||||
Version 3.3 | Version 3.2 | Version 3.0 | Version 2.0 | ||
Version 1.2 | Version 1.1 | Version 1.0 | |||
Version 0.99 | Version 0.96 | Version 0.95 | Version 0.9 | Version 0.8 | Version 0.7 |
print = 10
print print
An immediate core dump when setting a string element ≤ 0 or when setting a substring to anything else but a string has been fixed.
The size of the command history is now limited to 500 entries. This prevents the file "dpuser.history" to grow indefinitely.
Several pgplot routines could cause a core dump when called with inconsistent arguments. For those, the arguments supplied are checked.
The functions total, min, max, avg handle infinite and undefined values by ignoring them. Also, an optional value which should be ignored can be specified.
The handling of absolute file names under Microsoft Windows has been improved.
The functions min and max have been expanded in their functionality.
A few examples of the usage of DPUSER are provided in the examples/ subdirectory.
I have introduced a modulus operator ("%") applicable to integer arguments.
For the Win32 version of DPUSER, I was able to dig out a native port of rxvt for Windows. This offers many more features than the windpuser.exe program distributed with ealier versions. I slightly modified rxvt by adding a DPUSER icon.
The linux binary distribution is now statically linked against QT (as advertised in the Install section). This prevents an immediate core dump on RedHat linux.
Some memory leaks have been removed.
The median function now returns the single element value if applied to a matrix with only one element (instead of zero).
The nelements function now returns a value of one if not applied to a matrix but a number (either integer, double precision or complex).
Several bugs in reading huge FITS cubes (>256 MB) have been fixed. Now files up to a size of 2 GB can be read.
The rotate function used to multiply the total counts in an image by a factor of 4. This has been fixed.
The reass function had a bug when applied to matrices with an uneven axes length. This also affects the smooth function. Fixed.
The shift function has some new features.
In this version, some matrix arithmetics functions were added. This includes the "#" operator which does matrix multiplication.
Beginning with this version, it is possible to write user defined
procedures
and functions.
Some examples are provided in the dpuserlib/ directory.
!!!Feedback on this new feature is extremely welcome!!!
If a file named "startup.dpuser" is located either in your home directory or in a directory pointed to by the DPUSER environment variable, it will be executed during startup of DPUSER. An example is provided in the dpuserlib/ directory.
Some minor bugs (especially concerning arithmetics including double and single precision numbers) have been fixed.
The functions naxis, naxis1, naxis2, and naxis3 read in the values much faster when applied to FITS files (used to read in the complete file).
The ssa function is now a lot quicker when using the /resize switch.
A bug in the cubemedian function when applied to FITS files with NAXIS2 not a integer power of 2 has been fixed.
Subtracting a matrix from a real number resulted in wrong results. This also applies to complex numbers and FITS files. Along with this, a print -'hhh' didn't to the correct thing. This has been fixed.
Switch to cross-platform library QT. This affects strings, string arrays, as well as the dir and findfile functions. The interface doesn't change. This removed a memory leak in string arrays and fixes problems with findfile which would not list all files. The SFL library now is no longer needed to build DPUSER.
A new integer global variable "tmpmem" which gives the number of bytes which can be used for temporary storage. At the moment, this only affects the function cubemedian. The interface doesn't change.
A bug has been fixed in cubemerge which sometimes would not update the FITS header correctly.
The procedure sao accepts lower and upper limits; added /zscale and /log switch