CD Concept 6

home *** CD-ROM | disk | FTP | other *** search

/ CD Concept 6 / CD Concept 06.iso / mac / UTILITAIRE / RLaB / WHATS.NEW < prev next >

Wrap

Text File | 1995-01-25 | 19KB | 650 lines

*** What's New in Version v1.17 ? 1.) Added Inf capability to 2nd arg of norm (see help, obscure feature). 2.) hilbert.r: Hilbert transform, new file. 3.) Some other obscure features/changes in various plotting and numerical functions in order to make the testmatrix toolbox an easier port. 4.) Added frexp(), logb(), and log2(). 5.) Added second (optional) argument to getline to allow users to read in entire line as a string. 6.) Now you can do: </ v ; d /> = eig (a); This is be-grudgingly supplied to help Matlab users transition, and cut down on copies in functions. The thing on the left hand side is called an "open-list". See `help LIST' *** What's New in Version v1.15 ? (I think this should be 1.16) 1.) Some effort has been expended to re-work some of the guts in an attempt to safely speed interpreter execution. Additionally work has been done to clean up the parser generated op-codes, and simplify the functions (operations) in the virtual machine. Specifically: Re-work conditional interpretation. Speed up temporary entity handling. Due to the nature of the beast (RLaB), execution times will not change uniformly, some operations should be faster, some might not change much. But, on the whole, the trend should be faster execution (anywhere from 0-50% on my machine). 2.) Converted over to autoconf-v2.x. This should not mean anything to users, configure is still run in basically the same way. 3.) Implementation of standardized floating-point exception handling. Now rlab should ignore all floating point exceptions, except overflow. This works on SVr3.2 and SVR4.x, Suns, HPs, and Linux. If this gives you problems during build, then see the PROBLEMS file (run configure with `--disable-fpe'). 4.) A more accurate tic() and toc() using your system's times(2) function (if it exists). If your system does not have a times(2), then the older/less-precise tic() and toc() are used. Users should note that using tic() and toc(), in either implementation, is probably not valid for timing tests unless the test take at least a few seconds (> 10 is best). 5.) The scanner can now deal with Matlab style comments `%'. 6.) The rfile command is now more useful. Rfile handles multiple names and trailing comments. 7.) More improvements in the autoconfiguration. 8.) There is a new command, called `require' (see `help require'). This command is like rfile, except that it won't re-read files when the function already exists in the workspace. *** What's New in Version v1.13b ? 1.) There has been an internal overhaul of builtin function argument handling. No new features, just easier for people to write dynamically linked builtin functions (rex-files) once the documentation is done. 2.) fvscope now annotates non-function, global variables with an asterisk (`*'). 3.) New rfiles, section.r, show_prop.r. 4.) Added plsfile to allow users to specify Plplot output file. 5.) I have been working on configure again (a never ending process). Now using dynamic-linking is standard for SVr4 and Solaris. configure will look for plplot in /usr/local/plplot and /usr/local/... if nothing is specified. 6.) OVERHAUL FUNCTION SCOPING RULES. See `help FUNCTION', `help global', `help local', and `help static'. 7.) Functions can now use the variable `nargs' which is automatically initialized to the number of arguments passed to the function. see `help FUNCTION'. 8.) det() now returns zero after issuing a warning when given a singular matrix as an argument. 9.) ode() can now take an optional last argument that is a function to process the output. See `help ode'. 10.) The Primer has been updated to reflect new scoping rules. 11.) New gnuplot.r from Karl Storck. *** What's New in Version v1.09i ? 1.) Mostly a portability release. Binary I/O has been improved so it should work better on "non-standard" platforms (DEC-alpha). Some clean ups have been done to make RLaB build easier on HP-UX. 2.) `make install' has been overhauled to use version numbers on the executable, and the lib directory. The new install procedure will also try and move the old stuff out of the way. Nothing is removed, except the old executable. As always `make -n install' is highly recommended. 3.) Bug fixes to plhist, plhistx plotting. Addition of subplot() function. 4.) New builtin, putenv. 5.) New argument for plot, see `help plot'. Also some general bug fixes in the plotting routines. 6.) Add symmetric/Hermitian factorizations to `\', solve, factor, backsub. factor and solve now have an optional argument to allow the user to overide symmetry check. 7.) Bug fix in eval() string argument check. *** What's New in Version v1.09 ? 1.) Dynamic linking is functional for SVr4.0, SunOS-4.x, SunOS-5.x. However, there is no documentation at this point... Next release. *** What's New in Version v1.07 ? 1.) Not much except for some bug fixes (plugged a memory leak). Mostly there has been some re-organisation of the file and function structure in an attempt to make dynamic linking possible. Before we go any further though, a significant amount of testing should be done. *** What's New in Version v1.05 ? 1.) fread(), fseek(), if you are a C programmer you know what these are. The args are a little different. 2.) eval is now a built-in. Also eval returns the result of the last statement, and can be used in function with local variables. The best part is that eval no longer used temporary files to do its job. 3.) sign() is builtin. *** What's New in Version v1.02 ? 1.) readm() is MUCH faster. 2.) Revised toeplitz 3.) Added tmpfile() (builtin). 4.) Improved contour plotting so that only visible contours are displayed in legend. 5.) Improvements to `make Test' - much less use of rand(). *** What's New in Version v1.0 ? 1.) New functions (rfiles) bessely, besselj. 2.) local() modified so that arguments are copied. See `help FUNCTION' 3.) Numerically singlular inputs to solve(), \ are now handled with warnings and in a more uniform fashion. 4.) ./misc/rlab-mode.el I am NOT an Emacs-lisp programmer. This is hacked from tcl-mode.el. if anyone can improve it - go for it. 5.) Added plcont() - contour plotting. 6.) The primer has been updated to match version 1. *** What's New in Version 0.99i ? 1.) Added plhold(), plhold_off, and plegend(). 2.) New rfiles roots() and poly(), join() from T. S. Yang. 3.) A more recent version of fftpack. Sorry, but the accuracy is much better with the new version (rfft-1.2.tar.gz) - precision demands it. 4.) examples/plhold_demo.r is new. 5.) The plot help has been split up into separate files. *** What's New in Version 0.99f ? 1.) Added plmesh(). 2.) Added examples/curve_fit.r 3.) min(), max() now accept two numeric args. *** What's New in Version 0.99e ? 1.) Bug fixes. 2.) rfile will check the value of _rlab_search_path during execution. *** What's New in Version 0.99d ? 1.) Bug fixes. 2.) ode() argument list is now consistent with ode4() and ode78(). 3.) Added open() and getb(). *** What's New in Version 0.99 BETA ? 1.) New rfiles: mdsmax.r, and nmsmax.r. Both perform unconstrained optimization. 2.) Fixed non-Matlab-like behavior in acos(), and asin(). 3.) getline() returns a NULL string when a blank line is encountered. 4.) Added writeb() and readb(). Read and write binary files. MATLAB compatibility (see `help readb', `help writeb'). 5.) Added ode(). Ordinary differential equation integrator (netlib). 6.) Fixed readm() so it will read from pipes. 7.) Added getenv() built-in function. 8.) Plotting enhancements... plfont(), pltex(), plsori(), plwid(), plaxis(). 9.) filter() is now a builtin, compatible with MATLAB's implementation. 10.) Enhancements to matrix element-by-element operators, see `help OPERATORS'. Added `.+' and `.-' *** What's New in Version 0.97 BETA ? 1.) More plplot improvements. Updated primer to reflect plplot usage. 2.) Some non-Unix portability enhancements. 3.) Equality (== and !=) tests are allowed between numbers and strings. 4.) New builtin function schord(). See `help schord'. *** What's New in Version 0.96 BETA ? 1.) mod & atan2 functions now take various combinations of matrices. 2.) read() now taks optional 2nd argument (see `help read'). 3.) New fix(), and fixed up czt(). 4.) The configure script has been improved to take more options (see INSTALL). 5.) Changed eig (slightly), and added eigs (symmetric), eign (non-symmetric). Non-symmetric generalized eign() returns left-eigenvectors now. 6.) Added strtod(). 7.) Added PLPLOT built-in plotting capability. You must have the double precision version of the plplot library for it to work. See `help plplot' and `examples/plot_test.r'. 8.) Added sizeof() (see 'help sizeof') and re-worked whos(). *** What's New in Version 0.9(3-5) BETA ? 1.) New functions: complement(), intersection(), set(), union(), faxis(), fftplot(), window(), fmin(), dot(). 2.) readm(), and writem() have been modified so that the row and column values at the top of the file are no longer needed. This means old files written with writem() will have to have the first line filtered out. It also means that readm() should be able to directly read "flat-files" written by other programs. 3.) New & Improved qr(). New option qr(a,"p") for column pivoting. And qr() now works on MxN where M < N as well as M >= N. 4.) Function argument handling has been improved. my_func ( , , A ) is now possible. 5.) Many bug fixes. 6.) fft(), and ifft() now work like MATLAB's 7.) name() is gone (it wasn't of any use anymore anyway). 8.) Added lyap() (user-function), and sylv() (builtin-function). 9.) RLaB uses autoconf now. 10.) Added issymm(), see `help issymm'. 11.) Writem no longer closes the file after each call. 12.) New and improved eig, see `help eig'. 13.) Improved command line options. Added `-V' (version #), `-m' (no message on startup), and a usage message if an incorrect option is used. 14.) Added czt() - Chirp z-transform. *** What's New in Version 0.92 BETA ? 1.) New functions: input, num2str, funm, logm, disp, input, mret finite, printmat, int2str, filter, and isempty. 2.) A new builtin function called fvscope() (Function Variable SCOPE). Fvscope identifies the variables used in a function, and what their scope is. This is very useful for writing general purpose library functions. 3.) Bug fixes to the error recovery. *** What's New in Version 0.91 BETA ? 1.) RLaB can now load() files while executing code, such as in for, or while loops, or within functions. This makes a simple eval function possible. See `help load', `help eval', `help BUGS'. 2.) Added expm() rfile. *** What's New in Version 0.86 BETA ? 1.) Bug fixes. *** What's New in Version 0.85 BETA ? 1.) Enhanced svd(), See `help svd'. 2.) Bug fixes. *** What's New in Version 0.84 BETA ? 1.) Added schur() 2.) Bug fixes. *** What's New in Version 0.83 BETA ? 1.) The usual bug fixes. 2.) Added doc/tutorial.3.tex (List Tutorial). *** What's New in Version 0.82 BETA ? 1.) My apologies again, but I have restructured the class, type hierarchy in rlab. This was necessary for several reasons. The new hierarchy is only slightly different, and is described in the help files: class, type, and show. The new hierarchy looks like: (from the help file "type") CLASS: num TYPES: real complex CLASS: string TYPES: string CLASS: list TYPES: (see note 1.) CLASS: function TYPES: user builtin 2.) Improved type(). 3.) Improved test-suite. 4.) Improvements in the makefile install target. 5.) Rlab now has a command line editing option that is much smaller than GNU readline. Courtesy of Chris Thewalt. See `help COMMAND_EDIT' 6.) The contrib directory has been removed. Most of the contrib directories contents have been moved to the toolbox directory. I did this because I have been forgetting to keep the contrib directories files current with rlab. Also, the original intention was to keep unaltered, contributed rfiles there. Since I have had to modify the contents of contrib to keep up with the changes in rlab, the contrib directory no longer seemed appropriate. Maybe when rlab gets to v1.0. 7.) Small modifications to existing ODE integration functions. 8.) Added setterm(), showplot, phrd() to plot functionality. *** What's New in Version 0.78 BETA ? 1.) Added file static() declaration. 2.) Added exist() built-in function. 3.) diag(), ones() are builtin. 4.) Added whos(). 5.) Improved plot(), added pclose(). 6.) Added factor(), backsub(). lu() is now a user-function. 7.) My apologies, but I found the "dummy" name scheme for determining the status of a function argument to be totally inappropriate for nested function calls. Therefore, I have changed rlab behavior so that functions that are called with fewer args than declared are supplied UNDEFINED arguments to fill out the argument list. To test the status of an argument do: if (!exist (ARG)) { // initialize the ARG, error(), whatever... } 8.) Added qq_normal(), fliplr(), flipud() to ./toolbox *** What's New in Version 0.75 BETA ? 1.) Fixed cumsum(), and cumprod() (much more efficient). New rfiles save.r, clearall.r. 2.) Finished 1st draft of function tutorial. 3.) Improved getline(). 4.) Removed need to terminate function declaration with `;' or `\n'. 5.) The usual bug fixes, documentation improvements, etc... *** What's New in Version 0.73 BETA ? 1.) length() more compatible with the other length(). 2.) $$ can be used as global symbol table constant. *** What's New in Version 0.71 BETA ? 1.) cd(). New function. Changes current working directory. 2.) redit.r: For editing rfiles. 3.) Addition to config.h (HAVE_INDEX). *** What's New in Version 0.70 BETA ? 1.) Faster matrix multiplication functions ( ~ 20% ). 2.) sum(), and prod(), are builtin (faster). cumsum(), and cumprod() will be builtin shortly (rfiles for now. 3.) Better signal handling. 4.) Some interpreter speedups. 5.) Matrix powers finally work. 6.) Some type-checking/handling consistency improvements. 7.) The example plot() function is simpler now. 8.) Empty matrix syntax added ( `[]' ). 9.) fft(), ifft() now work on MxN matrices. max(), and min() any(), all(), sort(), work more like MATLAB versions. 10.) Many, many bugs fixed. 11.) `[ : ]' operation added. Forces Nx1 orientation of matrix. 12.) Complex atan, acos, asin, tan now work. 13.) A stop-gap matrix-tutorial in ./doc, updated man-page. 14.) .' non-conjugate transpose. 15.) getline() returns an empty list when nothing is read. See `help getline'. 16.) length(): New function. 17.) `#!' can be used in rlab scripts. *** What's New in Version 0.52 ALPHA ? 1.) Bug fixes. 2.) Now sort() returns a list of the sorted values, and the sorted indices. 3.) maxi()/mini() now return a scalar index value. This value can be used directly to index another matrix. *** What's New in Version 0.51 ALPHA ? 1.) The help and rfile commands are more versatile. For example: `help help' will work now. 2.) Vector creation (d1:d2:d3) rules make more sense (see `help VECTOR'). 3.) A\B, and C/D now use a more efficient algorithm when A or C is square. 4.) There is a brief man-page. 5.) Some bug fixes in max(), min(), and mod(). 6.) A directory of test-matrix functions has been added under the examples directory. *** What's New in Version 0.50 ALPHA ? 1.) The find function is now a builtin (much faster). 2.) The if-statement syntax has changed slightly. Used to be: if ( expression ) { statement(s) } else { // The braces and the else must be on the same line statement(s) } Now the correct syntax is: if ( expression ) { statement(s) else statement(s) } *** What's New in Version 0.32 ALPHA ? 1.) Added `rfile' command. `file name' now loads the file named `name.r' if it can be located in your RLAB_SEARCH_PATH. The simple command `rfile' prints a list of all the rfiles that appear in RLAB_SEARCH_PATH. 2.) Changed `help' command to be consistent with `rfile' command. The simple command `help' prints a list of: the help files in the default help directory, and the rfiles in RLAB_SEARCH_PATH. `help name' prints the content of the file `name' to the standard output. If the file does not exist in the default help directory, RLAB_SEARCH_PATH is searched. 3.) The `rlib' directory contents are now loaded upon startup, instead of having a HUGE `.rlab' file. Added RLAB_LIB_DIR for post-compile specification of library directory. 4.) All matrix output is paged. The pager is specified in the config.h file. 5.) Added getline() function. Works sort-of-like the awk getline. `x = getline("filename")' gets a line-at-time, and splits it into fields. Each field is a member of a list. Thus, x.[1] ... x.[N] contain the all the information from the line. 6.) Added strsplt(). Splits a string into a matrix of single characters. *** What's New in Version 0.30 ALPHA ? Version 0.3x is the last of the ALPHA test releases. Notice I did not say 0.30, since I do expect to release bug-fixes to v0.30. The major obstacle to starting beta test is a better online help, and a good printed manual. Other than bug-fixes these two goals will be my major priority. 1.) RLaB-0.30 is more MATLAB-like. 2.) Vectors have been remove as a distinct type. However, all of the usefulness of vectors is still accessible via single indexed matrices. Side effects: This means that the append operator, `,', does not function outside the [ ]. Instead, the comma (outside []) serves in a more traditional sense, as an argument list separator. Some of the built-in functions arguments are different. For example, zeros used to take a 2-element vector as an arg. Now zeros will accept two scalars, or a 2-element matrix. 3.) RLaB has adopted the MATLAB operator syntax. + - * / \ ^ Work in the traditional LA sense. .* ./ .\ .^ Work in the element-by-element sense. Note that `2./A' is the same as `2 ./ A' Also note that `^' is not fully implemented yet. 4.) The printed documentation has been removed. Two of the alpha testers have volunteered to overhaul the printed docs in latex or latexinfo. 5.) Some changes in the numeric output format, and the format() built-in function. See `help format'. 6.) RLaB will use matherr() instead of errno, If your machine supports it (and you select it in config.h). This option should prove much faster than checking errno after every math library function call. 7.) New built-in functions: inf(), nan(), round(). 8.) Matrix 2-norm has been added to norm(); `norm( m, "2" )'. 9.) "clock" option added to srand(); sets seed via the machine clock.