Mercurial > web-octave
annotate pages/NEWS-3.4.md @ 271:3ce19581b946
LICENSE: Add "GPL-3.0-or-later" for code
| author | Kai Torben Ohlhus <k.ohlhus@gmail.com> |
|---|---|
| date | Mon, 21 Feb 2022 17:32:38 +0900 |
| parents | e69093ab4992 |
| children |
| rev | line source |
|---|---|
| 215 | 1 --- |
| 2 layout: page | |
| 3 title: GNU Octave Version 3.4 | |
| 4 permalink: NEWS-3.4.html | |
| 5 --- | |
| 6 | |
| 7 ## Summary of important user-visible changes | |
| 8 | |
| 9 February 8, 2011 | |
| 10 | |
| 11 {% include release_news_select.md %} | |
| 12 | |
|
224
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
13 {::options parse_block_html="true" /} |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
14 <div class="panel callout"> |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
15 * TOC |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
16 {:toc} |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
17 </div> |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
18 {::options parse_block_html="false" /} |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
19 |
|
e69093ab4992
Improve Release Notes display.
Kai T. Ohlhus <k.ohlhus@gmail.com>
parents:
216
diff
changeset
|
20 |
| 215 | 21 <pre> |
| 22 ** IMPORTANT note about binary incompatibility in this release: | |
| 23 | |
| 24 Binary compatibility for all 3.4.x releases was originally planned, | |
| 25 but this is impossible for the 3.4.1 release due to a bug in the way | |
| 26 shared libraries were built in Octave 3.4.0. Because of this bug, | |
| 27 .oct files built for Octave 3.4.0 must be recompiled before they | |
| 28 will work with Octave 3.4.1. | |
| 29 | |
| 30 Given that there would be binary incompatibilities with shared | |
| 31 libraries going from Octave 3.4.0 to 3.4.1, the following | |
| 32 incompatible changes were also made in this release: | |
| 33 | |
| 34 * The Perl Compatible Regular Expression (PCRE) library is now | |
| 35 required to build Octave. | |
| 36 | |
| 37 * Octave's libraries and .oct files are now installed in | |
| 38 subdirectories of $libdir instead of $libexecdir. | |
| 39 | |
| 40 Any future Octave 3.4.x release versions should remain binary | |
| 41 compatible with Octave 3.4.1 as proper library versioning is now | |
| 42 being used as recommended by the libtool manual. | |
| 43 | |
| 44 ** The following functions have been deprecated in Octave 3.4.1 and will | |
| 45 be removed from Octave 3.8 (or whatever version is the second major | |
| 46 release after 3.4): | |
| 47 | |
| 48 cquad is_duplicate_entry perror strerror | |
| 49 | |
| 50 ** The following functions are new in 3.4.1: | |
| 51 | |
| 52 colstyle gmres iscolumn isrow mgorth nproc rectangle | |
| 53 | |
| 54 ** The get_forge_pkg function is now private. | |
| 55 | |
| 56 ** The rectangle_lw, rectangle_sw, triangle_lw, and triangle_sw | |
| 57 functions are now private. | |
| 58 | |
| 59 ** The logistic_regression_derivatives and logistic_regression_likelihood | |
| 60 functions are now private. | |
| 61 | |
| 62 ** ChangeLog files in the Octave sources are no longer maintained | |
| 63 by hand. Instead, there is a single ChangeLog file generated from | |
| 64 the Mercurial version control commit messages. Older ChangeLog | |
| 65 information can be found in the etc/OLD-ChangeLogs directory in the | |
| 66 source distribution. | |
| 67 | |
| 68 Summary of important user-visible changes for version 3.4: | |
| 69 --------------------------------------------------------- | |
| 70 | |
| 71 ** BLAS and LAPACK libraries are now required to build Octave. The | |
| 72 subset of the reference BLAS and LAPACK libraries has been removed | |
| 73 from the Octave sources. | |
| 74 | |
| 75 ** The `lookup' function was extended to be more useful for | |
| 76 general-purpose binary searching. Using this improvement, the | |
| 77 ismember function was rewritten for significantly better | |
| 78 performance. | |
| 79 | |
| 80 ** Real, integer and logical matrices, when used in indexing, will now | |
| 81 cache the internal index_vector value (zero-based indices) when | |
| 82 successfully used as indices, eliminating the conversion penalty for | |
| 83 subsequent indexing by the same matrix. In particular, this means it | |
| 84 is no longer needed to avoid repeated indexing by logical arrays | |
| 85 using find for performance reasons. | |
| 86 | |
| 87 ** Logical matrices are now treated more efficiently when used as | |
| 88 indices. Octave will keep the index as a logical mask unless the | |
| 89 ratio of true elements is small enough, using a specialized | |
| 90 code. Previously, all logical matrices were always first converted | |
| 91 to index vectors. This results in savings in both memory and | |
| 92 computing time. | |
| 93 | |
| 94 ** The `sub2ind' and `ind2sub' functions were reimplemented as compiled | |
| 95 functions for better performance. These functions are now faster, | |
| 96 can deliver more economized results for ranges, and can reuse the | |
| 97 index cache mechanism described in previous paragraph. | |
| 98 | |
| 99 ** The built-in function equivalents to associative operators (`plus', | |
| 100 `times', `mtimes', `and', and `or') have been extended to accept | |
| 101 multiple arguments. This is especially useful for summing | |
| 102 (multiplying, etc.) lists of objects (of possibly distinct types): | |
| 103 | |
| 104 matrix_sum = plus (matrix_list{:}); | |
| 105 | |
| 106 ** An FTP object type based on libcurl has been implemented. These | |
| 107 objects allow ftp connections, downloads and uploads to be | |
| 108 managed. For example, | |
| 109 | |
| 110 fp = ftp ("ftp.octave.org); | |
| 111 cd (fp, "gnu/octave"); | |
| 112 mget (fp, "octave-3.2.3.tar.bz2"); | |
| 113 close (fp); | |
| 114 | |
| 115 ** The default behavior of `assert (observed, expected)' has been | |
| 116 relaxed to employ less strict checking that does not require the | |
| 117 internals of the values to match. This avoids previously valid | |
| 118 tests from breaking due to new internal classes introduced in future | |
| 119 Octave versions. | |
| 120 | |
| 121 For instance, all of these assertions were true in Octave 3.0.x | |
| 122 but false in 3.2.x due to new optimizations and improvements: | |
| 123 | |
| 124 assert (2*linspace (1, 5, 5), 2*(1:5)) | |
| 125 assert (zeros (0, 0), []) | |
| 126 assert (2*ones (1, 5), (2) (ones (1,5))) | |
| 127 | |
| 128 ** The behavior of library functions `ismatrix', `issquare', and | |
| 129 `issymmetric' has been changed for better consistency. | |
| 130 | |
| 131 * The `ismatrix' function now returns true for all numeric, | |
| 132 logical and character 2-D or N-D matrices. Previously, `ismatrix' | |
| 133 returned false if the first or second dimension was zero. | |
| 134 Hence, `ismatrix ([])' was false, | |
| 135 while `ismatrix (zeros (1,2,0))' was true. | |
| 136 | |
| 137 * The `issquare' function now returns a logical scalar, and is | |
| 138 equivalent to the expression | |
| 139 | |
| 140 ismatrix (x) && ndims (x) == 2 && rows (x) == columns (x) | |
| 141 | |
| 142 The dimension is no longer returned. As a result, `issquare ([])' | |
| 143 now yields true. | |
| 144 | |
| 145 * The `issymmetric' function now checks for symmetry instead of | |
| 146 Hermitianness. For the latter, ishermitian was created. Also, | |
| 147 logical scalar is returned rather than the dimension, so | |
| 148 `issymmetric ([])' is now true. | |
| 149 | |
| 150 ** Function handles are now aware of overloaded functions. If a | |
| 151 function is overloaded, the handle determines at the time of its | |
| 152 reference which function to call. A non-overloaded version does not | |
| 153 need to exist. | |
| 154 | |
| 155 ** Overloading functions for built-in classes (double, int8, cell, | |
| 156 etc.) is now compatible with Matlab. | |
| 157 | |
| 158 ** Function handles can now be compared with the == and != operators, | |
| 159 as well as the `isequal' function. | |
| 160 | |
| 161 ** Performance of concatenation (using []) and the functions `cat', | |
| 162 `horzcat', and `vertcat' has been improved for multidimensional | |
| 163 arrays. | |
| 164 | |
| 165 ** The operation-assignment operators +=, -=, *= and /= now behave more | |
| 166 efficiently in certain cases. For instance, if M is a matrix and S a | |
| 167 scalar, then the statement | |
| 168 | |
| 169 M += S; | |
| 170 | |
| 171 will operate on M's data in-place if it is not shared by another | |
| 172 variable, usually increasing both time and memory efficiency. | |
| 173 | |
| 174 Only selected common combinations are affected, namely: | |
| 175 | |
| 176 matrix += matrix | |
| 177 matrix -= matrix | |
| 178 matrix .*= matrix | |
| 179 matrix ./= matrix | |
| 180 | |
| 181 matrix += scalar | |
| 182 matrix -= scalar | |
| 183 matrix *= scalar | |
| 184 matrix /= scalar | |
| 185 | |
| 186 logical matrix |= logical matrix | |
| 187 logical matrix &= logical matrix | |
| 188 | |
| 189 where matrix and scalar belong to the same class. The left-hand | |
| 190 side must be a simple variable reference. | |
| 191 | |
| 192 Moreover, when unary operators occur in expressions, Octave will | |
| 193 also try to do the operation in-place if it's argument is a | |
| 194 temporary expresssion. | |
| 195 | |
| 196 ** The effect of comparison operators (<, >, <=, and >=) applied to | |
| 197 complex numbers has changed to be consistent with the strict | |
| 198 ordering defined by the `max', `min', and `sort' functions. More | |
| 199 specifically, complex numbers are compared by lexicographical | |
| 200 comparison of the pairs `[abs(z), arg(z)]'. Previously, only real | |
| 201 parts were compared; this can be trivially achieved by wrapping the | |
| 202 operands in real(). | |
| 203 | |
| 204 ** The automatic simplification of complex computation results has | |
| 205 changed. Octave will now simplify any complex number with a zero | |
| 206 imaginary part or any complex matrix with all elements having zero | |
| 207 imaginary part to a real value. Previously, this was done only for | |
| 208 positive zeros. Note that the behavior of the complex function is | |
| 209 unchanged and it still produces a complex value even if the | |
| 210 imaginary part is zero. | |
| 211 | |
| 212 ** As a side effect of code refactoring in liboctave, the binary | |
| 213 logical operations are now more easily amenable to compiler | |
| 214 optimizations and are thus significantly faster. | |
| 215 | |
| 216 ** Octave now allows user-defined `subsasgn' methods to optimize out | |
| 217 redundant copies. For more information, see the manual. | |
| 218 | |
| 219 ** More efficient matrix division handling. Octave is now able to | |
| 220 handle the expressions | |
| 221 | |
| 222 M' \ V | |
| 223 M.' \ V | |
| 224 V / M | |
| 225 | |
| 226 (M is a matrix and V is a vector) more efficiently in certain cases. | |
| 227 In particular, if M is triangular, all three expressions will be | |
| 228 handled by a single call to xTRTRS (from LAPACK), with appropriate | |
| 229 flags. Previously, all three expressions required a physical | |
| 230 transpose of M. | |
| 231 | |
| 232 ** More efficient handling of certain mixed real-complex matrix | |
| 233 operations. For instance, if RM is a real matrix and CM a complex | |
| 234 matrix, | |
| 235 | |
| 236 RM * CM | |
| 237 | |
| 238 can now be evaluated either as | |
| 239 | |
| 240 complex (RM * real (CM), RM * imag (CM)) | |
| 241 | |
| 242 or as | |
| 243 | |
| 244 complex (RM) * CM, | |
| 245 | |
| 246 depending on the dimensions. The first form requires more | |
| 247 temporaries and copying, but halves the FLOP count, which normally | |
| 248 brings better performance if RM has enough rows. Previously, the | |
| 249 second form was always used. | |
| 250 | |
| 251 Matrix division is similarly affected. | |
| 252 | |
| 253 ** More efficient handling of triangular matrix factors returned from | |
| 254 factorizations. The functions for computing QR, LU and Cholesky | |
| 255 factorizations will now automatically return the triangular matrix | |
| 256 factors with proper internal matrix_type set, so that it won't need | |
| 257 to be computed when the matrix is used for division. | |
| 258 | |
| 259 ** The built-in `sum' function now handles the non-native summation | |
| 260 (i.e., double precision sum of single or integer inputs) more | |
| 261 efficiently, avoiding a temporary conversion of the whole input | |
| 262 array to doubles. Further, `sum' can now accept an extra option | |
| 263 argument, using a compensated summation algorithm rather than a | |
| 264 straightforward sum, which significantly improves precision if lots | |
| 265 of cancellation occurs in the summation. | |
| 266 | |
| 267 ** The built-in `bsxfun' function now uses optimized code for certain | |
| 268 cases where built-in operator handles are passed in. Namely, the | |
| 269 optimizations concern the operators `plus', `minus', `times', | |
| 270 `ldivide', `rdivide', `power', `and', `or' (for logical arrays), | |
| 271 the relational operators `eq', `ne', `lt', `le', `gt', `ge', and the | |
| 272 functions `min' and `max'. Optimizations only apply when both | |
| 273 operands are of the same built-in class. Mixed real/complex and | |
| 274 single/double operations will first convert both operands to a | |
| 275 common type. | |
| 276 | |
| 277 ** The `strfind' and `strrep' functions now have compiled | |
| 278 implementations, facilitating significantly more efficient searching | |
| 279 and replacing in strings, especially with longer patterns. The code | |
| 280 of `strcat' has been vectorized and is now much more efficient when | |
| 281 many strings are concatenated. The `strcmpi' and `strncmpi' | |
| 282 functions are now built-in functions, providing better performance. | |
| 283 | |
| 284 ** Matlab-style ignoring input and output function arguments using | |
| 285 tilde (~) is now supported. Ignored output arguments may be | |
| 286 detected from a function using the built-in function `isargout'. | |
| 287 For more details, consult the manual. | |
| 288 | |
| 289 ** The list datatype, deprecated since the introduction of cells, has | |
| 290 been removed. | |
| 291 | |
| 292 ** The accumarray function has been optimized and is now significantly | |
| 293 faster in certain important cases. | |
| 294 | |
| 295 ** The behavior of isreal and isnumeric functions was changed to be more | |
| 296 Matlab-compatible. | |
| 297 | |
| 298 ** The integer math & conversion warnings (Octave:int-convert-nan, | |
| 299 Octave:int-convert-non-int-val, Octave:int-convert-overflow, | |
| 300 Octave:int-math-overflow) have been removed. | |
| 301 | |
| 302 ** rem and mod are now built-in functions. They also handle integer | |
| 303 types efficiently using integer arithmetic. | |
| 304 | |
| 305 ** Sparse indexing and indexed assignment has been mostly rewritten. | |
| 306 Since Octave uses compressed column storage for sparse matrices, | |
| 307 major attention is devoted to operations manipulating whole columns. | |
| 308 Such operations are now significantly faster, as well as some other | |
| 309 important cases. | |
| 310 | |
| 311 Further, it is now possible to pre-allocate a sparse matrix and | |
| 312 subsequently fill it by assignments, provided they meet certain | |
| 313 conditions. For more information, consult the `spalloc' function, | |
| 314 which is no longer a mere dummy. Consequently, nzmax and nnz are no | |
| 315 longer always equal in Octave. Octave may also produce a matrix | |
| 316 with nnz < nzmax as a result of other operations, so you should | |
| 317 consistently use nnz unless you really want to use nzmax (i.e. the | |
| 318 space allocated for nonzero elements). | |
| 319 | |
| 320 Sparse concatenation is also affected, and concatenating sparse | |
| 321 matrices, especially larger collections, is now significantly more | |
| 322 efficient. This applies to both the [] operator and the | |
| 323 cat/vertcat/horzcat functions. | |
| 324 | |
| 325 ** It is now possible to optionally employ the xGESDD LAPACK drivers | |
| 326 for computing the singular value decomposition using svd(), instead | |
| 327 of the default xGESVD, using the configuration pseudo-variable | |
| 328 svd_driver. The xGESDD driver can be up to 6x times faster when | |
| 329 singular vectors are requested, but is reported to be somewhat less | |
| 330 robust on highly ill-conditioned matrices. | |
| 331 | |
| 332 ** Configuration pseudo-variables, such as page_screen_output or | |
| 333 confirm_recursive_rmdir (or the above mentioned svd_driver), now | |
| 334 accept a "local" option as second argument, requesting the change | |
| 335 to be undone when the current function returns: | |
| 336 | |
| 337 function [status, msg] = rm_rf (dir) | |
| 338 confirm_recursive_rmdir (false, "local"); | |
| 339 [status, msg] = rmdir (dir, "s"); | |
| 340 ... | |
| 341 endfunction | |
| 342 | |
| 343 Upon return, confirm_recursive_rmdir will be restored to the value | |
| 344 it had on entry to the function, even if there were subsequent | |
| 345 changes to the variable in function rm_rf or any of the functions | |
| 346 it calls. | |
| 347 | |
| 348 ** pkg now accepts a -forge option for downloading and installing | |
| 349 packages from Octave Forge automatically. For example, | |
| 350 | |
| 351 pkg install -forge general | |
| 352 | |
| 353 will automatically download the latest release of the general | |
| 354 package and attempt to install it. No automatic resolving of | |
| 355 dependencies is provided. Further, | |
| 356 | |
| 357 pkg list -forge | |
| 358 | |
| 359 can be used to list all available packages. | |
| 360 | |
| 361 ** The internal data representation of structs has been completely | |
| 362 rewritten to make certain optimizations feasible. The field data | |
| 363 can now be shared between structs with equal keys but different | |
| 364 dimensions or values, making operations that preserve the fields | |
| 365 faster. Economized storage is now used for scalar structs (just | |
| 366 like most other scalars), making their usage more | |
| 367 memory-efficient. Certain array-like operations on structs | |
| 368 (concatenation, uniform cellfun, num2cell) have gained a | |
| 369 significant speed-up. Additionally, the octave_scalar_map class | |
| 370 now provides a simpler interface to work with scalar structs within | |
| 371 a C++ DLD function. | |
| 372 | |
| 373 ** Two new formats are available for displaying numbers: | |
| 374 | |
| 375 format short eng | |
| 376 format long eng | |
| 377 | |
| 378 Both display numbers in engineering notation, i.e., mantissa + | |
| 379 exponent where the exponent is a multiple of 3. | |
| 380 | |
| 381 ** The following functions are new in Octave 3.4: | |
| 382 | |
| 383 accumdim erfcx nfields pqpnonneg uigetdir | |
| 384 bitpack fileread nth_element quadcc uigetfile | |
| 385 bitunpack fminbnd onCleanup randi uiputfile | |
| 386 blkmm fskipl pbaspect repelems uimenu | |
| 387 cbrt ifelse pie3 reset whitebg | |
| 388 curl ishermitian powerset rsf2csf | |
| 389 chop isindex ppder saveas | |
| 390 daspect luupdate ppint strread | |
| 391 divergence merge ppjumps textread | |
| 392 | |
| 393 ** Using the image function to view images with external programs such | |
| 394 as display, xv, and xloadimage is no longer supported. The | |
| 395 image_viewer function has also been removed. | |
| 396 | |
| 397 ** The behavior of struct assignments to non-struct values has been | |
| 398 changed. Previously, it was possible to overwrite an arbitrary | |
| 399 value: | |
| 400 | |
| 401 a = 1; | |
| 402 a.x = 2; | |
| 403 | |
| 404 This is no longer possible unless a is an empty matrix or cell | |
| 405 array. | |
| 406 | |
| 407 ** The dlmread function has been extended to allow specifying a custom | |
| 408 value for empty fields. | |
| 409 | |
| 410 ** The dlmread and dlmwrite functions have been modified to accept | |
| 411 file IDs (as returned by fopen) in addition to file names. | |
| 412 | |
| 413 ** Octave can now optimize away the interpreter overhead of an | |
| 414 anonymous function handle, if the function simply calls another | |
| 415 function or handle with some of its parameters bound to certain | |
| 416 values. Example: | |
| 417 | |
| 418 f = @(x) sum (x, 1); | |
| 419 | |
| 420 When f is called, the call is forwarded to @sum with the constant 1 | |
| 421 appended, and the anonymous function call does not occur on the | |
| 422 call stack. | |
| 423 | |
| 424 ** Deprecated functions. | |
| 425 | |
| 426 The following functions were deprecated in Octave 3.0 and have been | |
| 427 removed from Octave 3.4. | |
| 428 | |
| 429 beta_cdf geometric_pdf pascal_pdf | |
| 430 beta_inv geometric_rnd pascal_rnd | |
| 431 beta_pdf hypergeometric_cdf poisson_cdf | |
| 432 beta_rnd hypergeometric_inv poisson_inv | |
| 433 binomial_cdf hypergeometric_pdf poisson_pdf | |
| 434 binomial_inv hypergeometric_rnd poisson_rnd | |
| 435 binomial_pdf intersection polyinteg | |
| 436 binomial_rnd is_bool setstr | |
| 437 chisquare_cdf is_complex struct_contains | |
| 438 chisquare_inv is_list struct_elements | |
| 439 chisquare_pdf is_matrix t_cdf | |
| 440 chisquare_rnd is_scalar t_inv | |
| 441 clearplot is_square t_pdf | |
| 442 clg is_stream t_rnd | |
| 443 com2str is_struct uniform_cdf | |
| 444 exponential_cdf is_symmetric uniform_inv | |
| 445 exponential_inv is_vector uniform_pdf | |
| 446 exponential_pdf isstr uniform_rnd | |
| 447 exponential_rnd lognormal_cdf weibcdf | |
| 448 f_cdf lognormal_inv weibinv | |
| 449 f_inv lognormal_pdf weibpdf | |
| 450 f_pdf lognormal_rnd weibrnd | |
| 451 f_rnd meshdom weibull_cdf | |
| 452 gamma_cdf normal_cdf weibull_inv | |
| 453 gamma_inv normal_inv weibull_pdf | |
| 454 gamma_pdf normal_pdf weibull_rnd | |
| 455 gamma_rnd normal_rnd wiener_rnd | |
| 456 geometric_cdf pascal_cdf | |
| 457 geometric_inv pascal_inv | |
| 458 | |
| 459 The following functions were deprecated in Octave 3.2 and will | |
| 460 be removed from Octave 3.6 (or whatever version is the second major | |
| 461 release after 3.2): | |
| 462 | |
| 463 create_set spcholinv splu | |
| 464 dmult spcumprod spmax | |
| 465 iscommand spcumsum spmin | |
| 466 israwcommand spdet spprod | |
| 467 lchol spdiag spqr | |
| 468 loadimage spfind spsum | |
| 469 mark_as_command sphcat spsumsq | |
| 470 mark_as_rawcommand spinv spvcat | |
| 471 spatan2 spkron str2mat | |
| 472 spchol splchol unmark_command | |
| 473 spchol2inv split unmark_rawcommand | |
| 474 | |
| 475 The following functions have been deprecated in Octave 3.4 and will | |
| 476 be removed from Octave 3.8 (or whatever version is the second major | |
| 477 release after 3.4): | |
| 478 | |
| 479 autocor cellidx gammai krylovb values | |
| 480 autocov dispatch glpkmex replot | |
| 481 betai fstat is_global saveimage | |
| 482 | |
| 483 * For compatibility with Matlab, mu2lin (x) is now equivalent to | |
| 484 mu2lin (x, 0). | |
| 485 | |
| 486 * The ARPACK library is now distributed with Octave so it no longer | |
| 487 needs to be available as an external dependency when building | |
| 488 Octave. | |
| 489 </pre> |