forked from omnetpp/omnetpp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
WHATSNEW
7264 lines (5326 loc) · 318 KB
/
WHATSNEW
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
What's New
==========
This file summarizes OMNeT++ changes in each release. For changes related to
simulation model compatibility, see doc/API-Changes. For more detailed info
about all changes, see include/ChangeLog, src/*/ChangeLog, and ide/ChangeLog.
OMNeT++ 6.0pre15 (Dec 2021)
---------------------------
This prerelase contains minor fixes and cleanups in preparation to the final
6.0 release.
Build:
- The content of the `$PLATFORM` makefile variable has changed. Until know,
it contained the `platform.architecture` pair (e.g. `win32.x86_64`) which was
misleading. From now on, `Makefile.inc` provides `$PLATFORM` (e.g. `win32`)
and `$ARCH` (e.g. `x86_64`) in separate variables. `makefrag` files must be
updated if `$PLATFORM` variable was used in them.
- To build with AKAROA support, `WITH_AKAROA=yes` must be specified in the
`configure.user` file. The configuration script no longer tries to autodetect
it without explicitly requesting it.
IDE:
- Update to Eclipse 4.22, CDT 10.5 and PyDev 9.2. The private JRE bundled with
the IDE has been updated to Java 17 LTS.
Tools:
- `opp_run -v` now outputs the system architecture OMNeT++ was built for
(e.g. ARCH_X86_64 or ARCH_AARCH64).
- On Windows and macOS, the directory names in the `tools` directory have changed
to `win32.x86_64` and `macos.x86_64` respectively.
OMNeT++ 6.0pre14 (Nov 2021)
---------------------------
This prerelease, just about a month after the previous one, comes with two
important improvements in NED, follow-up changes in the simulation library,
and several improvements in Qtenv and the Analysis Tool in the IDE.
NED:
- In a complex model framework, there is usually a large number of (module or
channel) parameters, and so far it has not been obvious for users which
parameters can be meaningfully changed at runtime. For example, if a simple
module did not implement handleParameterChange() or the method did not cover
a particular nonvolatile parameter, then the user could technically change
that NED parameter at runtime, but the change did not take effect. This has
often caused confusion.
To remedy the situation, we introduced the @mutable property for parameters.
If a particular parameter supports changing its value at runtime (i.e. the
new value takes effect), then it should be marked with @mutable. If @mutable
is missing, then trying to change the parameter will result in a runtime
error ("Cannot change non-mutable parameter".)
Parameters of ned.DelayChannel and ned.DatarateChannel are now marked
@mutable.
To allow running existing simulation models that have not yet been updated
with @mutable annotations, we have added the parameter-mutability-check
configuration option. Setting parameter-mutability-check=false will give
back the old behavior.
- Added the expr() operator to NED, with the purpose of allowing models to
accept formulas or expressions which it could use e.g. to determine the
processing time of a packet based on packet length or other properties,
decide whether a packet should be allowed to pass or should be filtered out
based on its contents (filter condition), or to derive x and y coordinates
of a mobile node in the function of time.
The argument to expr() is a NED expression which will typically contain free
variables, (like x and y in expr(x+y)). The expr() operator creates an
object that encapsulates the expression, so it can be assigned to parameters
of the type "object". On the C++ side, the module implementation should
query the parameter to get at the expression object (of type
cOwnedDynamicExpression), and then it may bind the free variables and
evaluate the expression as often as it wishes.
- Object parameters now allow simple values (int, double, bool, string) as
well. C++-wise, they are stored in the parameter in cValueHolder-wrapped
cValue objects.
MSG:
- The @readonly property was added.
Core:
- cPar: Implemented isMutable() and the mechanism behind the new @mutable
property.
- Added the cOwnedDynamicExpression class which holds the result of NED expr()
operators. cOwnedDynamicExpression is both cOwnedObject and
cDynamicExpression (multiple inheritance). To make this possible, the
cObject base class was removed from cExpression.
- Added cValueHolder class. It instantiated as wrapper object around cValue
when a non-object value (double, string, etc) is assigned to a NED parameter
of the type "object".
- Introduced cValueContainer as common base class for cValueHolder,
cValueArray and cValueMap.
- cValueMap/cValueArray: revert dup() to behave the same way cArray/cQueue
does, i.e. only those objects that are owned by the cloned container are
duplicated. This normally doesn't make much difference, but some NED
expressions that relied on the old behavior now need to be adjusted. For
example:
object foo = {};
object bar = dup([foo,foo]); // --> error (contents of foo is not
// cloned by cloning the array)
object baz = [dup(foo), dup(foo)] // OK
- getModuleByPath() / findModuleByPath() moved to cComponent, so that
they can be called on cChannels too.
- Internal classes, global variables, etc moved into the omnetpp::internal
namespace.
Qtenv:
- Log / Messages view: Slight change in the notation used for the source and
destination modules of the message, in order to make it unambiguous. Use
explicit "." and "^" to indicate the location of the module. Also, it now
uses arrows of uniform lengths everywhere.
- Added the "Fira Code" font as embedded resource, and set it to be used by
default for the log window. The reason is that it provides nice --> arrow
ligatures in the Messages view.
- Fixed a bug where all further EV_* logging output was discarded after any
module printed something into the log in its constructor.
Analysis Tool:
- On the Browse Data page, mask values that don't fit into their columns using
'####'. This is akin to what MS Excel does.
- Improved the appearance of Matplotlib-emulated native charts (which are used
if a native chart is exported from the IDE or opp_charttool): added digit
grouping to tick labels, made its appearance more consistent with the actual
native plot regarding axis limits, number format for ticks, etc.
- In the Python API, restored the omnetpp.scave.plot module (which was renamed
ideplot in the last prerelease), in order not to break existing ANF files.
Importing it will delegate to ideplot after issuing a "this is deprecated"
warning.
- Numerous bugs were fixed, including a complete IDE crash in rare cases.
OMNeT++ 6.0pre13 (Oct 2021)
---------------------------
Core:
- Introduced cEnvir::pausePoint()
- cValue changed to hold any_ptr instead of cObject. This change involves
several changes, e.g. type OBJECT renamed to POINTER, and pointerValue() added.
- cClassDescriptor: Added getFieldValue()/setFieldValue() methods
to allow accessing fields in a typed way, using cValue. Previously
existing methods getFieldValueAsString()/setFieldValueAsString()
only allowed string-based access.
- Message compiler: Follow cClassDescriptor change: generate getFieldValue()
and setFieldValue() methods. Added new properties @toValue()/@fromValue()
so that user can define how to implement those.
- cPar: added cValue-based generic access: getValue(), setValue()
- cProperty: added getNumKeys() method; updateWith() made public
- cResultFilter, cResultRecorder: grouped init() args into a Context struct,
The old init() methods have been preserved as deprecated (and invoked from
the new init()) in case an existing filter/recorder overrides them. Note
that potential external calls to the old init() method won't work any more
(they will have no effect), and need to be changed to the new version.
- Added MergeFilter, a result filter that allows multiple inputs,
and multiplexes them onto its singe output. It is available (as the
merge() function) in the source= part of @statistic.
- Fixed histogram loading issue in the output scalar file (.sca). Bin edges
that are very close could become equal in the file if insufficient printing
precision was set, rendering the file unreadable. The issue is now handled
both during result file writing (if such condition is detected, bin edges
are written with full [16-digit] precision) and reading (zero-width bins are
merged into adjacent nonzero-width bin).
- Further bug fixes and minor improvements.
Qtenv:
- Object Inspector: Added possibility to select display mode (Children/
Grouped/Flat/Inheritance) per-node. If display mode override is specified
for a node, it will affect the whole subtree. The display mode for the
selected tree can be changed via hotkey (Ctrl+B) or via the context menu.
- Log: Show component init stages as banners, but hide empty ones.
- Prevent manually enabled "Debug on Errors" setting to be turned
off by (lack of) configuration option.
- Various bug fixes.
IDE:
- Analysis Tool: Multiple UI improvements:
- When plotting results from the Browse Data page, prefer native plotting
to Matplotlib. Native plots are both more scalable and also arguably
more convenient to use.
- In the Scatter Plot config dialog, the "Iso line" terminology changed to
"Group by" (suggested by Edward Longman).
- When setting the name of a chart, also set its title, if it was empty before.
- Improve focus handling on tab switching in the Chart Properties dialog.
- Fix the bug where middle-click closing a chart tab would discard changes silently.
- Added "timewinthruput" vector operation
- Python API: The "plot" module (omnetpp.scave.plot) was renamed to "ideplot".
Charts in existing ANF files need to be either recreated, or the import
line and usages of the plot module in the chart script need to be changed
accordingly.
- Added support for ARM-based Linux distros
- JustJ: Use stripped version of JDK 15 instead of JDK 11
Build:
- `setenv` is used now on Windows to set up the build environment.
It is automatically sourced when mingwenv.cmd or vcenv.cmd is started.
- If `setenv_local` file is present in the installation root, it is
automatically sourced from the `setenv` script. This can be used to
set up user specific environment variables.
Documentation:
- Install Guide updated. New sections in the macOS chapter: "Enabling
Development Mode in the Terminal", "Running OMNeT++ on Apple Silicon"
OMNeT++ 6.0pre12 (Jul 2021)
---------------------------
Simulation kernel:
- NED functions that take or return values of type 'object' are now allowed.
- cModule now keeps track of submodule vectors as entities, and not just as
a collection of submodules with vector indices, meaning that we can now
distinguish between nonexistent and zero-size submodule vectors. Random
access of vector elements has also became more efficient (constant-time
operation). Several new methods have been added as part of this change:
hasSubmoduleVector(), getSubmoduleVectorSize(), addSubmoduleVector(),
deleteSubmoduleVector(), setSubmoduleVectorSize(), getSubmoduleVectorNames().
- As part of the above change, several cModule and cModuleType methods have been
added or updated. A partial list:
- In cModule, the hasSubmodule(), getSubmoduleNames(), hasGateVector(),
hasGates() methods have been added for consistency between submodule and
gate APIs.
- The return type of the getGateNames() method has been changed from
std::string<const char*> to std::vector<std::string>, for consistency
with getSubmoduleNames().
- cModule: added setIndex() and setNameAndIndex().
- cModule, cGate: getIndex(), getVectorSize(), gateSize() and similar methods
now throw for non-vector submodules/gates.
- cModule: add separate addGateVector() method instead misusing addGate()
for creating gate vectors, also for consistency with addSubmoduleVector().
- In cModuleType::create(), the vectorSize parameter of create() has been removed
- In cModuleType::createScheduleInit(), an index argument was added to allow
creating submodule vector elements.
- cModule: added addSubmodule() method which flows more naturally than
cModuleType::create().
- There have been changes in submodule and channel iterators. SubmoduleIterator
has been rewritten due to the change in how submodule vectors are represented,
which may affect the iteration order in some cases. Iterators now throw an
exception if a change occurs in the list of submodules/channels during
iteration. Their APIs have also changed a little: operator-- was removed,
and init(m) renamed to reset().
- Improvements in the multi-stage initialization protocol with regard to
dynamic module creation. Modules now keep track of the last init stage
they completed (lastCompletedInitStage). During an init stage, iteration
of modules is restarted if creation/deletion is detected during iteration;
modules already initialized in the previous round are recognized and skipped
with the help of lastCompletedInitStage.
- cComponent: added getNedTypeAndFullName() and getNedTypeAndFullPath().
They are especially useful in constructing error messages in NED functions.
- Minor changes in the model change notification classes. In
cPreModuleAddNotification, the vectorSize field was removed (as it was
redundant), and in cPostModuleDeleteNotification, the interpretation of
the index field has changed a little: if the deleted module was not part
of a module vector, index is now set to -1 instead of 0.
- New classes: opp_pooledstring, opp_staticpooledstring. They provide pool-
backed string storage (reference counted and non reference counted,
respectively). In turn, the cStringPool class was removed; use either
of the pooled string classes or or opp_staticpooledstring::get(const char *)
instead.
- cValueArray: added asIntVectorInUnit(), asDoubleVectorInUnit() functions
- Store origin (file:line) in cPar parameters (more precisely, in
cDynamicExpression), so we can report it on evaluation errors.
Most visible change: cPar::parse() gained an extra FileLine argument.
Also, cDynamicExpression now has get/setSourceLocation()
- cNedFunction: added search by name AND accepted number of args.
- cXMLElement change: resolve $MODULE_INDEX, $PARENTMODULE_NAME and similar
variables to empty string if they are not applicable.
- Several string functions have been made available for models. A
representative partial list: opp_isempty(), opp_isblank(),
opp_nulltoempty(), opp_trim(), opp_split(), opp_splitandtrim(),
opp_join(), opp_stringendswith(), opp_substringbefore(), etc.
- cStringTokenizer has been rewritten. It now supports features like
optional skipping of empty tokens, optional trimming of tokens, optional
honoring of quotes, optional honoring of parens/brackets/brackets
(i.e. the input string is not broken into tokens in the middle of a
parenthesized expression).
- cClassDescriptor: methods changed to use any_ptr instead of void*
for passing the object. (any_ptr is a smart pointer class that provides
type safety for void* pointers.) Pointers need to be put into and extracted
from any_ptr using the new toAnyPtr() / fromAnyPtr() functions. They have
specialized versions for each type (via templates and overloading).
For new types, the message compiler generates toAnyPtr()/fromAnyPtr()
in the header file. For the simulation library classes, these methods
come from "sim_std_m.h" (generated from sim_std.msg); "sim_std_m.h"
is now part of <omnetpp.h>.
- Various other improvements in the simulation library.
NED:
- Semantics change: Within a submodule's or connection's curly brace block,
the interpretation of parameter/gate/submodule references that don't use an
explicit "parent" or "this" qualifier has changed. They no longer refer to
the enclosing compound module's item, but to the item of (or within) the
local submodule or channel object. The interpretation of item references
outside subcomponent blocks has remained unchanged. An example:
network Network {
parameters:
int foo;
submodules:
node1: { ... }
node2: Node {
foo = foo; // ERROR: self-reference! Change to 'foo=parent.foo'.
bar = this.foo; // OK
baz = node1.foo; // ERROR: refers to yet-uncreated submodule node2.node1!
bax = parent.node1.foo; // OK
}
}
The set of forms accepted by exists() and sizeof() has been expanded.
Notably, exists() can now be used to check the existence of an item in a
submodule vector (as submodule vectors may now contain holes, due to
@omittedTypename). However, the index() syntax has been removed, index is
only accepted without parens.
Note that referencing a submodule's submodule from NED (i.e. the a.b or
this.a.b syntax within a submodule block) is generally wrong, because
submodules are only created after parameters have already been set up. (An
exception is volatile parameters, as they are only evaluated when the
parameter value is actually read.)
Error messages have been revised and expanded to be more helpful and
facilitate porting NED code from 5.x.
To write NED which is compatible with both OMNeT++ version 5.x and 6.0,
qualify all references within subcomponent blocks with explicit "parent"
or "this", and require OMNeT++ 5.7 which is the first and only 5.x version
that understands the "parent" keyword.
- Added support for Universal Function Call Syntax (UFCS) to NED, which means
that now any function call may also be written as if the function was a
method of its first argument (provided it has one). That is, f(x,..) can now
be also written in the alternative form x.f(...). This results in improved
readability in certain cases, and allows chaining function calls.
- New NED functions: get(), size(), eval(), dup(). eval() evaluates a NED
expression given as string. dup() simply calls an object's C++ dup() method;
it is useful because module parameters of type 'object' only accept objects
they fully own, which can be worked around using dup(). An example:
object a = [1,2,3];
object b = dup(a); // without dup() it is an error: value is owned by parameter 'a'
- NED functions for reading (from file) and parsing (from string) of CSV,
JSON and XML files: parseCSV(), readCSV(), parseExtendedCSV(),
readExtendedCSV(), parseJSON(), readJSON(), parseExtendedJSON(),
readExtendedJSON(), parseXML(), readXML(). The "extended" variants support
expressions in the file, which will be evaluated during parsing. The XML
functions are aliases to the existing xml()/xmldoc() functions. Bonus
file-related functions: readFile(), workingDir(), baseDir(), resolveFile(),
absFilePath().
- Allow the body of a parametric submodule to assign apparently nonexistent
parameters. Example:
app: <default("FooApp")> like IApp {
parameters:
address = parent.address;
sleepTime = 1s;
}
If FooApp has a sleepTime parameter but IApp does not, then running with
the defaults is fine, but changing to another IApp type that does NOT
have a sleepTime parameter will stop the simulation with an error.
Solution: inside a parametric submodule type, we simply ignore parameter
assignments that assign seemingly nonexistent parameters.
- Implemented @omittedTypename property. @omittedTypename allows one to
specify a NED type to use when typename="" is specified for a parametric
submodule or channel, or to remove (skip creating) that submodule, submodule
vector element, or channel altogether.
@omittedTypename can be specified on a module interface or channel
interface, or on a submodule or connection. It should contain a single
(optional) value. The value names the type to use. If it is absent, the
submodule or channel will not be created. (The connection will be created
without a channel object.)
- Improved error reporting during network building: when an error occurs
during assigning a parameter from a NED file, the error message now includes
the location of the assignment (file:line in NED file).
Ini files:
- It is now possible to break long lines without using a trailing backslash.
Continuation lines are marked as such by indenting them, i.e. an indented
line is now interpreted as a continuation of the previous line. (It is not
possible to break a line inside a string constant that way.) Breaking lines
using a trailing backslash way still works (and it can also be used
to break string constants, too).
- The "Config " prefix in section headers is now optional. That is,
[Config PureAloha] may now be also written as [PureAloha], the two
are equivalent.
IDE:
- Several improvements in various editors.
- Update to Eclipse 4.20, CDT 10.3 and PyDev 8.3. Another change is that
the JRE in the IDE is now provided by Eclipse JustJ project.
OMNeT++ 6.0pre11 (Apr 2021)
---------------------------
This prerelease contains many small changes and improvements. The highlight
of this release is the Sequence Chart tool in the IDE which has been revised
and significantly extended with new functionality. NOTE: the eventlog (.elog)
file format has changed in a non-backward-compatible way.
The summary of changes since 6.0pre10:
Core:
- Added display name support to modules. Display name is a string that
optionally appears in Qtenv next to (or instead of) the normal module
name, in order to help the user distinguish between similarly-named
submodules. For example, application-layer modules app[0], app[1], etc.
in INET may be given descriptive names like "file transfer", "video",
"voice", or "CBR", and have them displayed in Qtenv. Display names may be
set using the **.display-name per-module configuration option, or
programmatically by calling setDisplayName().
- Added the "g=<group>" (layout group) display string tag, which makes it
possible to apply predefined arrangements like row, column, matrix or ring
to a group of unrelated submodules. This layouting feature was previously
only available for submodule vectors. When "g" tags are used, submodules
in the same group are now regarded for layouting purposes as if they
were part of the same submodule vector.
- Made it possible to specify display strings in the configuration. The
value given in the **.display-string per-component configuration option
is merged into the component's display string in the same way inheritance
or submodule display strings work: it may add, overwrite or remove items
from it.
- Change in the parameter change notification mechanism: Calls to the
handleParameterChange() method during initialization are no longer
suppressed. Because now every change results in a notification, the
umbrella handleParameter(nullptr) call at the end of the initialization
is no longer needed and has been removed. The consequence is that
handleParameterChange() methods need to be implemented more carefully,
because they may be called at a time when the module may not have
completed all initialization stages. Also, if an existing model relied on
handleParameter(nullptr) being called, it needs to be updated.
- Module name and full name (i.e. name[index]) are now stringpooled, which
reduces memory usage in exchange for a small build-time extra cost.
In addition, string pools now use std::unordered_map instead of std::map,
which results in improved performance.
- Signature change of the cVisitor::visit(cObject *obj) virtual method: it
can now request end of iteration via returning false (hence, return type
changed from void to bool) instead of throwing EndTraversalException.
Existing cVisitor subclasses in model code will need to be adjusted.
- cValue's doubleValue() and intValue() methods now throw an exception when
called on a value that has a measurement unit, in order to reduce usage
mistakes. If the value has a unit, call either doubleValueInUnit() /
intValueInUnit(), or doubleValueRaw()/intValueRaw() plus getUnit().
- The setPerformFinalGC() method was removed. It was meant for internal
use, and pretty much unused by model code.
- cHasher: add(const std::string&) changed to be consistent with the
add(const char *) overload. This may cause fingerprint changes in models
that use it.
- Eventlog recording: Implemented snapshot and incremental index support.
This introduces means significant (breaking) changes in the elog file
format. Several new configuration options to control eventlog recording:
eventlog-max-size, eventlog-min-truncated-size, eventlog-snapshot-frequency,
eventlog-index-frequency, eventlog-options. See also changes in the
Sequence Chart Tool below.
- Many other bug fixes and improvements; see the ChangeLog files under src/
for details.
Qtenv:
- Added support for showing a submodule's display name under the icon
instead of, or in addition to, the normal name. The format can be
selected in the context menu.
- Added support for new columns in the packet log view: TxUpdate, Durations,
Length, Info. This was implemented in the cDefaultMessagePrinter class
which is part of the simulation library. This change makes it possible
to see if a packet is a transmission update (except if the simulation
installs its own packet printer like INET does).
Eventlog:
- The eventlog file format has been changed substantially. You won't be able
to open older eventlog files with this version of the IDE, nor new files in
previous IDE versions.
- The most prominent change is the introduction of snapshot and index chunks in
addition to the already recorded event chunks. The former contains a complete
snapshot of the relevant simulation state without referring to any other line
in the file. The latter provides the set of changes in the relevant simulation
state since the last index or snapshot chunk. The contents of the index chunk
is expressed as references to other lines of the eventlog file. This change in
the eventlog format provides the foundation for a better user experience in
the IDE especially for large eventlog files.
- Several new eventlog recording related configuration options have been added
to the INI file format.
- the 'eventlog-snapshot-frequency' configuration option specifies how often
snapshots are recorded in the evenlog file. Snapshots help various tools to
handle large eventlog files more efficiently. Specifying greater value means
less help, while smaller value means bigger eventlog files.
- the 'eventlog-index-frequency' configuration option specifies how often
indices are recorder in the eventlog file. An index is much smaller than a
full snapshot, but it only contains the differences since the last index.
Specifying greater value means less help, while smaller value means bigger
eventlog files.
- the 'eventlog-max-size' configuration option specifies the maximum size of
the eventlog file in bytes. The eventlog file is automatically truncated
when this limit is reached.
- the 'eventlog-min-truncated-size' configuration options specifies the minimum
size of the eventlog file in bytes after the file is truncated. Truncation
means older events are discarded while newer ones are kept.
- the 'eventlog-options' configuration option allows to record only certain
categories reducing the eventlog file size (e.g. text, message, module,
methodcall, displaystring).
Sequence Chart:
- The visualization of events have been horizontally expanded on the chart, now
it uses long rounded rectangles instead of small circles. This change allows
the separate visualization of method calls without overlapping. If method
calls are hidden then the chart falls back to the old behavior where events
are small circles.
- The set of displayed axes can be explicitly configured independently from the
set of displayed events. The chart now also supports displaying events on a
compound module axis if the corresponding submodule axis is hidden. Collapsing
and expanding module axes can be done using the mouse.
- The chart now also displays a set of axis headers on the left side that shows
the module hierarchy in a compact and interactive way.
- All parts of the sequence chart can be separately shown/hidden.
- The selection has been extended with support for axis and method calls.
- A new time measurement feature has been added that allows measuring the time
difference between multiple events and other interesting points in time.
- Several automatic configuration presets have been added (e.g. network level,
full detail, default) to make the configuration for the most common chart use
cases easier.
- A new pattern matching based user defined colorization (e.g. message sends,
method calls, axes, events) feature has been added. This feature allows
creating more easily understandable sequence charts by encoding certain model
properties as colors.
- The chart can display eventlog file statistics and operation statistics.
Analysis Tool:
- Allow editing the pages of the "Configure Chart" dialog. The "Edit pages"
button in the dialog brings up the new "Edit Chart Dialog Pages" dialog.
- Chart template gallery improvements: improved dialog and content.
- "Reset Chart To Template": added dialog for choosing which elements to reset.
- Properties view to show chart properties when chart is selected.
- Vector Operations: on selecting the operation from the context menu,
ask the filter's parameters before adding it.
- In the "Configure Chart" dialog, added content assist to the "Vector
Operations" field.
- Make native plot legend background and border semi-transparent.
- Automatically refresh open chart when set of loaded result files change.
- Chart page toolbar: added "Refresh Result Files" icon; zoom icons redesigned;
other improvements.
- Improved scatterchart and boxplots chart templates.
- Fix: Don't move the cursor upon error when the source editor is focused.
- Many more bugfixes and improvements; see the ChangeLog files under ui/
for details.
Inifile Editor:
- Added support for ignorable options. User can mark custom config options
as ones to be ignored by the editor.
- Get rid of asynchronous/delayed form page reread in FormPage.
IDE General:
- Install Simulation Models dialog: Added support for changing the project
name and location on download. The (default) project name is now coming
from the descriptor file even if INET is installed on the Welcome screen.
- Turned on showing line numbers ruler in source editors by default.
- Added an XML editor, language server, terminal, and marketplace client
to the IDE.
Tools:
- opp_charttool: The functionality of opp_charttool has been made available
for Python programs as a library module called "omnetpp.scave.analysis".
OMNeT++ 6.0pre10 (December 2020)
--------------------------------
This prerelease contains many small changes and improvements. Additionally,
due to improvements in the toolchain and the build process, Windows users
may see the linking time of large models like INET Framework to drop
dramatically (1-2 orders magnitude, e.g. from several minutes to several
seconds).
The summary of changes since 6.0pre9:
Core:
- Refactoring: Some classes, methods and variables related to ownership
management were renamed: cDefaultOwner -> cSoftOwner; defaultOwner ->
owningContext, etc.
- Slight change to the operation of preDelete() introduced in the previous
prerelease: it is now called via callPreDelete() which does the recursion part,
so preDelete() itself doesn't recurse any more. Also, we switch the context
to the module for the duration of the preDelete() call.
- Two utility functions were added to cObject: getClassAndFullPath() and
getClassAndFullName(). They are mostly useful in logging and error messages.
- Canvas API: Added text alignment support to text and label figures:
cFigure::Alignment enum, getAlignment()/setAlignment() in cAbstractTextFigure.
- Canvas API: cFigure::RGBA: added toAlpha() method, and a constructor taking
Color.
- Fingerprints: Due a bugfix in cHasher::add(const char *), fingerprints that
involve hashing strings changed their values.
- Fingerprints: Changed the way fingerprints are computed from figures. Most
importantly, fingerprints are now affected by all visual properties, not just
geometry information. This change only affects fingerprints that contain the
'f' (=figures) ingredient.
- Changed the configuration options that control the operation of the "Fake GUI"
Cmdenv feature, which is useful for computing fingerprints from simulations
with custom animation. See config options starting with "cmdenv-fakegui-".
- Removed the support for OMNeT++ 4.x fingerprints (USE_OMNETPP4x_FINGERPRINTS).
- Changed the NED package exclusion mechanism that was introduced in an earlier
6.0 prerelease. Exclusions are now specified in terms of package names
instead of folders. To this end, the "-x" command-line option changed meaning
(takes package list instead of folder list), and also, the ned-exclusion-path
config option was renamed to ned-package-exclusions, and the OMNETPP_NED_-
EXCLUSION_PATH environment variable to OMNETPP_NED_PACKAGE_EXCLUSIONS.
- Source code modernization: use in-class member initializers wherever possible.
Qtenv:
- Turn off animating method calls by default. The rationale is that method
call animations usually expose low-level (C++ implementation level)
details on the GUI, which are rarely of interest to a casual user.
- Added the possibility to disable method calls locally (per module type)
via the context menu, even when the global switch in the Preferences dialog
is turned on.
- Changed the factory default of "Allow backward arrows for hops" to false.
- Fix osgEarth viewpoints ignoring SRS (PR #851).
IDE:
- Numerous bug fixes
- Launcher: Pass NED exclusions to the simulation program as -x options.
NED exclusions typically come from disabled project features, e.g. in INET.
Build:
- Allow using the LLD linker by default (if present on the system). LLD is
potentially much faster than the standard GNU linker. For example, the
use of LLD, together with other changes that were part done in INET,
reduced INET linking time on Windows from 380s (!) to 2s (!!!) in DEBUG
mode, and from 90s to 5s in RELEASE mode on one of our dev boxes.
- When building DLL files (shared libraries) on Windows, the build process now
generates <name>.def and <name>.dll.a files in addition to the resulting
<name>.dll file. The <name>.ddl.a file is a static import library that can
be used by other dependent projects to link with the model in the shared
library. The <name>.def file can be used to check what methods were actually
exported from the model.
Misc:
- Install Guide updated.
OMNeT++ 6.0pre9 (October 2020)
------------------------------
A long time, six months have passed since the last prerelease, and, as a
consequence, we have accumulated a lot of changes. Arguably the most important
change is an update to the OMNeT++ programming model: the introduction of the
possibility to update ongoing packet transmissions, that is, the ability to
abort, shorten, or extend a packet (usually an L2 frame) while it is being
transmitted. Significant amount of work has been put into improving the looks
and functionality of Qtenv, and into refining the Python-based Analysis Tool in
the IDE. There were numerous other improvements as well -- read on for details.
Transmission updates in the simulation kernel:
- A description of the new "transmission updates" functionality follows.
We start with an overview, then go into API details and related changes.
- The initial send() is interpreted as: "packet transmission begins now,
packet content and duration are, as things are now, going to be this".
Following that, an "update" (or any number of updates) can be sent.
An update is a packet with the updated ("actual") content, and with a
"remaining transmission duration" attached. Updates may only
be sent while transmission is still ongoing.
As an example, aborting a transmission is done by sending a
packet with a truncated content and a remaining duration of zero.
Transmission updates are paired with the original packet they modify
using a transmissionId. The transmissionId is normally chosen to be the
packet ID of the original transmission. Channels should understand
updates and handle them accordingly.
Receivers that receive the packet at the end of the reception, which is the
default operating mode, will only receive the final update. The original
packet and intermediate updates are absorbed by the simulation kernel.
Receivers that receive the packet at the start of the reception (see
cGate::setDeliverImmediately(), ex setDeliverOnReceptionStart())
should be prepared to receive all of the original packet and the
updates, and handle them appropriately. Tx updates can be recognized
from cPacket::isUpdate() returning true. cPacket::getRemainingDuration()
returns the remaining transmission duration, and cPacket::getDuration()
the total transmission duration.
As a safeguard against unprepared modules accidentally processing
tx updates as normal full-blown packets, the module is only given
tx updates if it explicitly declares that it is able to handle them
properly. The latter is done by the module calling
setTxUpdateSupport(true) e.g. in initialize().
Non-transmission channels don't handle updates any different from
other messages and packets.
Details and related changes follow.
- send() and sendDirect() now accept a SendOptions struct where optional
parameters such as delay can be passed in. sendDelayed() and other
send()/sendDirect() variants now convert their extra args to a SendOptions,
and delegate to the "standard" send()/sendDirect() versions. SendOptions was
introduced as a means to handle combinatorial explosion of send() variants.
- cDatarateChannel now allows overriding the duration (which is normally
computed from the packet length and the channel datarate) from the send()
call.
- cDatarateChannel: non-packet messages now pass through without
interfering with packets.
- cDatarateChannel: disabled channels now let transmission updates through,
so that it is possible for the transmitter module to abort the
ongoing packet transmission.
- Note: tx updates (without duration/remainingDuration) are allowed
on paths without transmission channels.
- Signature change in methods that participate in the send()/sendDirect()
protocol: SendOptions was added to the arg list. This affects:
cGate::deliver(), cModule::arrived(), cChannel::processMessage()
- In cChannel::processMessage(), result_t was renamed cChannel::Result,
and it is now a proper return value (not an output parameter)
- remainingDuration was added to cChannel::Result.
- cDatarateChannel's datarate is now optional: set it to 0 or nan
to leave it unspecified. This change was necessary to support
transmitting frames with per-frame data rate selection.
- cDatarateChannel now optionally allows multiple concurrent transmissions,
with or without any bookkeeping and associated checks. This is useful for
modeling a channel with contention-based access and collisions, such as
classic Ethernet. The operating mode has to be selected programmatically,
with the channel's setMode() method. Possible modes are SINGLE, MULTI and
UNCHECKED.
- The forceTransmissionFinishTime() method of channels has been deprecated.
It was always meant as a temporary device to allow implementing aborting
frame transmissions, and now with the arrival of the new transmission
update API there is no reason to use it any more. Simulations using it
should be migrated to the new API.
- Rename: setDeliverOnReceptionStart() -> setDeliverImmediately()
- Added cSimpleModule::supportsTxUpdates() flag
- cPacket now carries a remainingDuration field
- cPacket: eliminated FL_ISRECEPTIONSTART; isReceptionStart() now uses
remainingDuration as input; added a similar isReceptionEnd() method.
- cPacket::str() overhaul to reflect new fields and uses
- In the APIs, sendDelay and propagationDelay (which were sort of
combined into a single value) are now distinct, and handled
separately.
- Eventlog format changes: BeginSendEntry: added sendDelay and
transmissionId; SendHopEntry, SendDirectEntry: added remainingDuration
Simulation kernel, other changes:
- cSimpleModule: added scheduleAfter(), rescheduleAt(), rescheduleAfter()
methods. This is mainly for convenience, but using rescheduleAt() instead of
a cancelEvent() followed by a scheduleAt() will eventually allow for a more
efficient implementation.
- getModuleByPath() adjustment: It will now never return nullptr, even for
the empty string given as path. Instead it will throw an exception if the
module was not found. This change makes this method consistent with other
getter methods in the simulation library, and allows nullptr checks to be
removed from model code that uses it.
- cComponent: added preDelete(). This a virtual method that the user can
override to add tasks to be done before the module (or channel) is deleted.
When deleteModule() is called on a compound module, it first invokes
preDelete() recursively on the submodule tree, and only starts deleting
modules after that. preDelete() can help simplify network or module deletion
in a complex simulation that involves model change listeners.
- Added opp_component_ptr<T>. It implements a smart pointer that points to
a cComponent (i.e. a module or channel), and automatically becomes nullptr
when the referenced object is deleted. It is a non-owning ("weak") pointer,
i.e. the pointer going out of scope has no effect on the referenced object.
opp_component_ptr<T> can be useful in implementing modules that hold
pointers to other modules and want to be prepared for those modules
getting deleted. It can also be useful for simplifying safe destruction
of compound modules containing such modules.
- A new method, findModuleByPath() was added for cases when an optionally
existing module needs to be found.
- cIListener's destructor now unsubscribes from all places it was subscribed
to. This change was necessitated by the following deleteModule() change.
- deleteModule(): Module destruction sequence was changed so that when
deleting a compound module, the compound module's local listeners are
notified about the deletion of the submodules.
- deleteModule() internals refactored. The motivation was to avoid doing
things like firing pre-model-change notifications from a halfway-deleted
module. Now we do every potentially risky thing (such as deleting submodules
and disconnecting gates) from doDeleteModule(), and only delete the module
object when it is already barebones (no submodules, gates, listeners, etc).
With this change, the deletion sequence is now pretty much the reverse of
the setup sequence.
- Now it is allowed for modules to be deleted (including self-deletion) and
created at will during initialization.
- cMessage: allow isMessage() to be called from subclasses.
- Added the autoWarmupFilter statistic attribute that allows one to disable
auto-adding the warmup filter to a statistic. Example:
@statistic[foo](record=vector;autoWarmupFilter=false);
This will cause all values from the foo signal to be recorded, even
values emitted during the warm-up period (if one is set).
However, the real motivation behind this feature is to allow the user to add
the warm-up filter at a non-default location in the filter chain. By
default, the warm-up filter is added at the start of the filter chain (right
before the first filter), but this is not always what is needed.
For example, "@statistic[foo](source=min;record=vector)" is equivalent to
"@statistic[foo](source=min(warmup(foo));autoWarmupFilter=false;record=vector)",
and records (as vector) the minimum of the values which were emitted after
the warmup period is over. In contrast, if we replace "min(warmup(foo))"
with "warmup(min(foo))", it will compute the minimum of ALL values, but only
starts recording the result after the warmup period has expired.
This is a crucial difference sometimes. For example, a statistic might
record queue length computed as the difference of the number of messages
that entered the queue and those that left it:
@statistic[queueLen](source=count(pkIn)-count(pkOut);record=vector); //INCORRECT
In this case, if a warmup period is set, the result may even go negative
because the pkIn and pkOut signals that arrive during the warmup period
are ignored, and if pkOut arrives after that, we are at -1. The correct
solution is to add the warmup filter after the difference between arrivals
and departures have been computed:
@statistic[queueLen](source=warmup(count(pkIn)-count(pkOut));autoWarmupFilter=false;record=vector); //OK
The autoWarmupFilter option exists because either location for the warmup
filter (beginning or end of the chain, or even mid-chain) may make sense for
certain statistics. The model author needs to decide per-statistic which one
is correct.
- Added DemuxFilter, which allows splitting the stream of values arriving
from a signal to multiple streams. For example, if values from the foo