Mercurial > web-octave
comparison NEWS-3.4.html @ 94:e8fc61e077fc
Merged closed branch "kai" into default.
author | Kai T. Ohlhus <k.ohlhus@gmail.com> |
---|---|
date | Tue, 01 Nov 2016 01:06:10 +0100 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
92:7609e2a6faef | 94:e8fc61e077fc |
---|---|
1 <pre> | |
2 Summary of important user-visible changes for version 3.4.3: | |
3 ----------------------------------------------------------- | |
4 | |
5 ** Octave 3.4.3 is a bug fixing release. | |
6 | |
7 Summary of important user-visible changes for version 3.4.2: | |
8 ----------------------------------------------------------- | |
9 | |
10 ** Octave 3.2.4 fixes some minor installation problems that affected | |
11 version 3.4.1. | |
12 | |
13 Summary of important user-visible changes for version 3.4.1: | |
14 ----------------------------------------------------------- | |
15 | |
16 ** Octave 3.4.1 is primarily a bug fixing release. | |
17 | |
18 ** IMPORTANT note about binary incompatibility in this release: | |
19 | |
20 Binary compatibility for all 3.4.x releases was originally planned, | |
21 but this is impossible for the 3.4.1 release due to a bug in the way | |
22 shared libraries were built in Octave 3.4.0. Because of this bug, | |
23 .oct files built for Octave 3.4.0 must be recompiled before they | |
24 will work with Octave 3.4.1. | |
25 | |
26 Given that there would be binary incompatibilities with shared | |
27 libraries going from Octave 3.4.0 to 3.4.1, the following | |
28 incompatible changes were also made in this release: | |
29 | |
30 * The Perl Compatible Regular Expression (PCRE) library is now | |
31 required to build Octave. | |
32 | |
33 * Octave's libraries and .oct files are now installed in | |
34 subdirectories of $libdir instead of $libexecdir. | |
35 | |
36 Any future Octave 3.4.x release versions should remain binary | |
37 compatible with Octave 3.4.1 as proper library versioning is now | |
38 being used as recommended by the libtool manual. | |
39 | |
40 ** The following functions have been deprecated in Octave 3.4.1 and will | |
41 be removed from Octave 3.8 (or whatever version is the second major | |
42 release after 3.4): | |
43 | |
44 cquad is_duplicate_entry perror strerror | |
45 | |
46 ** The following functions are new in 3.4.1: | |
47 | |
48 colstyle gmres iscolumn isrow mgorth nproc rectangle | |
49 | |
50 ** The get_forge_pkg function is now private. | |
51 | |
52 ** The rectangle_lw, rectangle_sw, triangle_lw, and triangle_sw | |
53 functions are now private. | |
54 | |
55 ** The logistic_regression_derivatives and logistic_regression_likelihood | |
56 functions are now private. | |
57 | |
58 ** ChangeLog files in the Octave sources are no longer maintained | |
59 by hand. Instead, there is a single ChangeLog file generated from | |
60 the Mercurial version control commit messages. Older ChangeLog | |
61 information can be found in the etc/OLD-ChangeLogs directory in the | |
62 source distribution. | |
63 | |
64 Summary of important user-visible changes for version 3.4: | |
65 --------------------------------------------------------- | |
66 | |
67 ** BLAS and LAPACK libraries are now required to build Octave. The | |
68 subset of the reference BLAS and LAPACK libraries has been removed | |
69 from the Octave sources. | |
70 | |
71 ** The `lookup' function was extended to be more useful for | |
72 general-purpose binary searching. Using this improvement, the | |
73 ismember function was rewritten for significantly better | |
74 performance. | |
75 | |
76 ** Real, integer and logical matrices, when used in indexing, will now | |
77 cache the internal index_vector value (zero-based indices) when | |
78 successfully used as indices, eliminating the conversion penalty for | |
79 subsequent indexing by the same matrix. In particular, this means it | |
80 is no longer needed to avoid repeated indexing by logical arrays | |
81 using find for performance reasons. | |
82 | |
83 ** Logical matrices are now treated more efficiently when used as | |
84 indices. Octave will keep the index as a logical mask unless the | |
85 ratio of true elements is small enough, using a specialized | |
86 code. Previously, all logical matrices were always first converted | |
87 to index vectors. This results in savings in both memory and | |
88 computing time. | |
89 | |
90 ** The `sub2ind' and `ind2sub' functions were reimplemented as compiled | |
91 functions for better performance. These functions are now faster, | |
92 can deliver more economized results for ranges, and can reuse the | |
93 index cache mechanism described in previous paragraph. | |
94 | |
95 ** The built-in function equivalents to associative operators (`plus', | |
96 `times', `mtimes', `and', and `or') have been extended to accept | |
97 multiple arguments. This is especially useful for summing | |
98 (multiplying, etc.) lists of objects (of possibly distinct types): | |
99 | |
100 matrix_sum = plus (matrix_list{:}); | |
101 | |
102 ** An FTP object type based on libcurl has been implemented. These | |
103 objects allow ftp connections, downloads and uploads to be | |
104 managed. For example, | |
105 | |
106 fp = ftp ("ftp.octave.org); | |
107 cd (fp, "gnu/octave"); | |
108 mget (fp, "octave-3.2.3.tar.bz2"); | |
109 close (fp); | |
110 | |
111 ** The default behavior of `assert (observed, expected)' has been | |
112 relaxed to employ less strict checking that does not require the | |
113 internals of the values to match. This avoids previously valid | |
114 tests from breaking due to new internal classes introduced in future | |
115 Octave versions. | |
116 | |
117 For instance, all of these assertions were true in Octave 3.0.x | |
118 but false in 3.2.x due to new optimizations and improvements: | |
119 | |
120 assert (2*linspace (1, 5, 5), 2*(1:5)) | |
121 assert (zeros (0, 0), []) | |
122 assert (2*ones (1, 5), (2) (ones (1,5))) | |
123 | |
124 ** The behavior of library functions `ismatrix', `issquare', and | |
125 `issymmetric' has been changed for better consistency. | |
126 | |
127 * The `ismatrix' function now returns true for all numeric, | |
128 logical and character 2-D or N-D matrices. Previously, `ismatrix' | |
129 returned false if the first or second dimension was zero. | |
130 Hence, `ismatrix ([])' was false, | |
131 while `ismatrix (zeros (1,2,0))' was true. | |
132 | |
133 * The `issquare' function now returns a logical scalar, and is | |
134 equivalent to the expression | |
135 | |
136 ismatrix (x) && ndims (x) == 2 && rows (x) == columns (x) | |
137 | |
138 The dimension is no longer returned. As a result, `issquare ([])' | |
139 now yields true. | |
140 | |
141 * The `issymmetric' function now checks for symmetry instead of | |
142 Hermitianness. For the latter, ishermitian was created. Also, | |
143 logical scalar is returned rather than the dimension, so | |
144 `issymmetric ([])' is now true. | |
145 | |
146 ** Function handles are now aware of overloaded functions. If a | |
147 function is overloaded, the handle determines at the time of its | |
148 reference which function to call. A non-overloaded version does not | |
149 need to exist. | |
150 | |
151 ** Overloading functions for built-in classes (double, int8, cell, | |
152 etc.) is now compatible with Matlab. | |
153 | |
154 ** Function handles can now be compared with the == and != operators, | |
155 as well as the `isequal' function. | |
156 | |
157 ** Performance of concatenation (using []) and the functions `cat', | |
158 `horzcat', and `vertcat' has been improved for multidimensional | |
159 arrays. | |
160 | |
161 ** The operation-assignment operators +=, -=, *= and /= now behave more | |
162 efficiently in certain cases. For instance, if M is a matrix and S a | |
163 scalar, then the statement | |
164 | |
165 M += S; | |
166 | |
167 will operate on M's data in-place if it is not shared by another | |
168 variable, usually increasing both time and memory efficiency. | |
169 | |
170 Only selected common combinations are affected, namely: | |
171 | |
172 matrix += matrix | |
173 matrix -= matrix | |
174 matrix .*= matrix | |
175 matrix ./= matrix | |
176 | |
177 matrix += scalar | |
178 matrix -= scalar | |
179 matrix *= scalar | |
180 matrix /= scalar | |
181 | |
182 logical matrix |= logical matrix | |
183 logical matrix &= logical matrix | |
184 | |
185 where matrix and scalar belong to the same class. The left-hand | |
186 side must be a simple variable reference. | |
187 | |
188 Moreover, when unary operators occur in expressions, Octave will | |
189 also try to do the operation in-place if it's argument is a | |
190 temporary expresssion. | |
191 | |
192 ** The effect of comparison operators (<, >, <=, and >=) applied to | |
193 complex numbers has changed to be consistent with the strict | |
194 ordering defined by the `max', `min', and `sort' functions. More | |
195 specifically, complex numbers are compared by lexicographical | |
196 comparison of the pairs `[abs(z), arg(z)]'. Previously, only real | |
197 parts were compared; this can be trivially achieved by wrapping the | |
198 operands in real(). | |
199 | |
200 ** The automatic simplification of complex computation results has | |
201 changed. Octave will now simplify any complex number with a zero | |
202 imaginary part or any complex matrix with all elements having zero | |
203 imaginary part to a real value. Previously, this was done only for | |
204 positive zeros. Note that the behavior of the complex function is | |
205 unchanged and it still produces a complex value even if the | |
206 imaginary part is zero. | |
207 | |
208 ** As a side effect of code refactoring in liboctave, the binary | |
209 logical operations are now more easily amenable to compiler | |
210 optimizations and are thus significantly faster. | |
211 | |
212 ** Octave now allows user-defined `subsasgn' methods to optimize out | |
213 redundant copies. For more information, see the manual. | |
214 | |
215 ** More efficient matrix division handling. Octave is now able to | |
216 handle the expressions | |
217 | |
218 M' \ V | |
219 M.' \ V | |
220 V / M | |
221 | |
222 (M is a matrix and V is a vector) more efficiently in certain cases. | |
223 In particular, if M is triangular, all three expressions will be | |
224 handled by a single call to xTRTRS (from LAPACK), with appropriate | |
225 flags. Previously, all three expressions required a physical | |
226 transpose of M. | |
227 | |
228 ** More efficient handling of certain mixed real-complex matrix | |
229 operations. For instance, if RM is a real matrix and CM a complex | |
230 matrix, | |
231 | |
232 RM * CM | |
233 | |
234 can now be evaluated either as | |
235 | |
236 complex (RM * real (CM), RM * imag (CM)) | |
237 | |
238 or as | |
239 | |
240 complex (RM) * CM, | |
241 | |
242 depending on the dimensions. The first form requires more | |
243 temporaries and copying, but halves the FLOP count, which normally | |
244 brings better performance if RM has enough rows. Previously, the | |
245 second form was always used. | |
246 | |
247 Matrix division is similarly affected. | |
248 | |
249 ** More efficient handling of triangular matrix factors returned from | |
250 factorizations. The functions for computing QR, LU and Cholesky | |
251 factorizations will now automatically return the triangular matrix | |
252 factors with proper internal matrix_type set, so that it won't need | |
253 to be computed when the matrix is used for division. | |
254 | |
255 ** The built-in `sum' function now handles the non-native summation | |
256 (i.e., double precision sum of single or integer inputs) more | |
257 efficiently, avoiding a temporary conversion of the whole input | |
258 array to doubles. Further, `sum' can now accept an extra option | |
259 argument, using a compensated summation algorithm rather than a | |
260 straightforward sum, which significantly improves precision if lots | |
261 of cancellation occurs in the summation. | |
262 | |
263 ** The built-in `bsxfun' function now uses optimized code for certain | |
264 cases where built-in operator handles are passed in. Namely, the | |
265 optimizations concern the operators `plus', `minus', `times', | |
266 `ldivide', `rdivide', `power', `and', `or' (for logical arrays), | |
267 the relational operators `eq', `ne', `lt', `le', `gt', `ge', and the | |
268 functions `min' and `max'. Optimizations only apply when both | |
269 operands are of the same built-in class. Mixed real/complex and | |
270 single/double operations will first convert both operands to a | |
271 common type. | |
272 | |
273 ** The `strfind' and `strrep' functions now have compiled | |
274 implementations, facilitating significantly more efficient searching | |
275 and replacing in strings, especially with longer patterns. The code | |
276 of `strcat' has been vectorized and is now much more efficient when | |
277 many strings are concatenated. The `strcmpi' and `strncmpi' | |
278 functions are now built-in functions, providing better performance. | |
279 | |
280 ** Matlab-style ignoring input and output function arguments using | |
281 tilde (~) is now supported. Ignored output arguments may be | |
282 detected from a function using the built-in function `isargout'. | |
283 For more details, consult the manual. | |
284 | |
285 ** The list datatype, deprecated since the introduction of cells, has | |
286 been removed. | |
287 | |
288 ** The accumarray function has been optimized and is now significantly | |
289 faster in certain important cases. | |
290 | |
291 ** The behavior of isreal and isnumeric functions was changed to be more | |
292 Matlab-compatible. | |
293 | |
294 ** The integer math & conversion warnings (Octave:int-convert-nan, | |
295 Octave:int-convert-non-int-val, Octave:int-convert-overflow, | |
296 Octave:int-math-overflow) have been removed. | |
297 | |
298 ** rem and mod are now built-in functions. They also handle integer | |
299 types efficiently using integer arithmetic. | |
300 | |
301 ** Sparse indexing and indexed assignment has been mostly rewritten. | |
302 Since Octave uses compressed column storage for sparse matrices, | |
303 major attention is devoted to operations manipulating whole columns. | |
304 Such operations are now significantly faster, as well as some other | |
305 important cases. | |
306 | |
307 Further, it is now possible to pre-allocate a sparse matrix and | |
308 subsequently fill it by assignments, provided they meet certain | |
309 conditions. For more information, consult the `spalloc' function, | |
310 which is no longer a mere dummy. Consequently, nzmax and nnz are no | |
311 longer always equal in Octave. Octave may also produce a matrix | |
312 with nnz < nzmax as a result of other operations, so you should | |
313 consistently use nnz unless you really want to use nzmax (i.e. the | |
314 space allocated for nonzero elements). | |
315 | |
316 Sparse concatenation is also affected, and concatenating sparse | |
317 matrices, especially larger collections, is now significantly more | |
318 efficient. This applies to both the [] operator and the | |
319 cat/vertcat/horzcat functions. | |
320 | |
321 ** It is now possible to optionally employ the xGESDD LAPACK drivers | |
322 for computing the singular value decomposition using svd(), instead | |
323 of the default xGESVD, using the configuration pseudo-variable | |
324 svd_driver. The xGESDD driver can be up to 6x times faster when | |
325 singular vectors are requested, but is reported to be somewhat less | |
326 robust on highly ill-conditioned matrices. | |
327 | |
328 ** Configuration pseudo-variables, such as page_screen_output or | |
329 confirm_recursive_rmdir (or the above mentioned svd_driver), now | |
330 accept a "local" option as second argument, requesting the change | |
331 to be undone when the current function returns: | |
332 | |
333 function [status, msg] = rm_rf (dir) | |
334 confirm_recursive_rmdir (false, "local"); | |
335 [status, msg] = rmdir (dir, "s"); | |
336 ... | |
337 endfunction | |
338 | |
339 Upon return, confirm_recursive_rmdir will be restored to the value | |
340 it had on entry to the function, even if there were subsequent | |
341 changes to the variable in function rm_rf or any of the functions | |
342 it calls. | |
343 | |
344 ** pkg now accepts a -forge option for downloading and installing | |
345 packages from Octave Forge automatically. For example, | |
346 | |
347 pkg install -forge general | |
348 | |
349 will automatically download the latest release of the general | |
350 package and attempt to install it. No automatic resolving of | |
351 dependencies is provided. Further, | |
352 | |
353 pkg list -forge | |
354 | |
355 can be used to list all available packages. | |
356 | |
357 ** The internal data representation of structs has been completely | |
358 rewritten to make certain optimizations feasible. The field data | |
359 can now be shared between structs with equal keys but different | |
360 dimensions or values, making operations that preserve the fields | |
361 faster. Economized storage is now used for scalar structs (just | |
362 like most other scalars), making their usage more | |
363 memory-efficient. Certain array-like operations on structs | |
364 (concatenation, uniform cellfun, num2cell) have gained a | |
365 significant speed-up. Additionally, the octave_scalar_map class | |
366 now provides a simpler interface to work with scalar structs within | |
367 a C++ DLD function. | |
368 | |
369 ** Two new formats are available for displaying numbers: | |
370 | |
371 format short eng | |
372 format long eng | |
373 | |
374 Both display numbers in engineering notation, i.e., mantissa + | |
375 exponent where the exponent is a multiple of 3. | |
376 | |
377 ** The following functions are new in Octave 3.4: | |
378 | |
379 accumdim erfcx nfields pqpnonneg uigetdir | |
380 bitpack fileread nth_element quadcc uigetfile | |
381 bitunpack fminbnd onCleanup randi uiputfile | |
382 blkmm fskipl pbaspect repelems uimenu | |
383 cbrt ifelse pie3 reset whitebg | |
384 curl ishermitian powerset rsf2csf | |
385 chop isindex ppder saveas | |
386 daspect luupdate ppint strread | |
387 divergence merge ppjumps textread | |
388 | |
389 ** Using the image function to view images with external programs such | |
390 as display, xv, and xloadimage is no longer supported. The | |
391 image_viewer function has also been removed. | |
392 | |
393 ** The behavior of struct assignments to non-struct values has been | |
394 changed. Previously, it was possible to overwrite an arbitrary | |
395 value: | |
396 | |
397 a = 1; | |
398 a.x = 2; | |
399 | |
400 This is no longer possible unless a is an empty matrix or cell | |
401 array. | |
402 | |
403 ** The dlmread function has been extended to allow specifying a custom | |
404 value for empty fields. | |
405 | |
406 ** The dlmread and dlmwrite functions have been modified to accept | |
407 file IDs (as returned by fopen) in addition to file names. | |
408 | |
409 ** Octave can now optimize away the interpreter overhead of an | |
410 anonymous function handle, if the function simply calls another | |
411 function or handle with some of its parameters bound to certain | |
412 values. Example: | |
413 | |
414 f = @(x) sum (x, 1); | |
415 | |
416 When f is called, the call is forwarded to @sum with the constant 1 | |
417 appended, and the anonymous function call does not occur on the | |
418 call stack. | |
419 | |
420 ** Deprecated functions. | |
421 | |
422 The following functions were deprecated in Octave 3.0 and have been | |
423 removed from Octave 3.4. | |
424 | |
425 beta_cdf geometric_pdf pascal_pdf | |
426 beta_inv geometric_rnd pascal_rnd | |
427 beta_pdf hypergeometric_cdf poisson_cdf | |
428 beta_rnd hypergeometric_inv poisson_inv | |
429 binomial_cdf hypergeometric_pdf poisson_pdf | |
430 binomial_inv hypergeometric_rnd poisson_rnd | |
431 binomial_pdf intersection polyinteg | |
432 binomial_rnd is_bool setstr | |
433 chisquare_cdf is_complex struct_contains | |
434 chisquare_inv is_list struct_elements | |
435 chisquare_pdf is_matrix t_cdf | |
436 chisquare_rnd is_scalar t_inv | |
437 clearplot is_square t_pdf | |
438 clg is_stream t_rnd | |
439 com2str is_struct uniform_cdf | |
440 exponential_cdf is_symmetric uniform_inv | |
441 exponential_inv is_vector uniform_pdf | |
442 exponential_pdf isstr uniform_rnd | |
443 exponential_rnd lognormal_cdf weibcdf | |
444 f_cdf lognormal_inv weibinv | |
445 f_inv lognormal_pdf weibpdf | |
446 f_pdf lognormal_rnd weibrnd | |
447 f_rnd meshdom weibull_cdf | |
448 gamma_cdf normal_cdf weibull_inv | |
449 gamma_inv normal_inv weibull_pdf | |
450 gamma_pdf normal_pdf weibull_rnd | |
451 gamma_rnd normal_rnd wiener_rnd | |
452 geometric_cdf pascal_cdf | |
453 geometric_inv pascal_inv | |
454 | |
455 The following functions were deprecated in Octave 3.2 and will | |
456 be removed from Octave 3.6 (or whatever version is the second major | |
457 release after 3.2): | |
458 | |
459 create_set spcholinv splu | |
460 dmult spcumprod spmax | |
461 iscommand spcumsum spmin | |
462 israwcommand spdet spprod | |
463 lchol spdiag spqr | |
464 loadimage spfind spsum | |
465 mark_as_command sphcat spsumsq | |
466 mark_as_rawcommand spinv spvcat | |
467 spatan2 spkron str2mat | |
468 spchol splchol unmark_command | |
469 spchol2inv split unmark_rawcommand | |
470 | |
471 The following functions have been deprecated in Octave 3.4 and will | |
472 be removed from Octave 3.8 (or whatever version is the second major | |
473 release after 3.4): | |
474 | |
475 autocor cellidx gammai krylovb values | |
476 autocov dispatch glpkmex replot | |
477 betai fstat is_global saveimage | |
478 | |
479 * For compatibility with Matlab, mu2lin (x) is now equivalent to | |
480 mu2lin (x, 0). | |
481 | |
482 * The ARPACK library is now distributed with Octave so it no longer | |
483 needs to be available as an external dependency when building | |
484 Octave. | |
485 </pre> |