1 SimGrid (3.4-svn) unstable; urgency=high
4 * Tested successfully on AIX
7 * fix a bug on struct sizeof computation, which prevented the
8 exchange of arrays of structs in some conditions
9 - added a regression test about this in datadesc_usage
10 * Allow the exchange of 0-long dynamic vectors.
11 - for that, use -1 as indicator of dynamic size instead of 0
12 - This implied to change any size from unsigned long to long,
13 reducing a bit communication abilities, but I guess that with
14 64bits being quite common, this is more than enough.
15 - This also induce a protocol change, thus bumping network protocol
16 version from 0 to 1 (if we have external users, we have to get
17 clean on that point too ;)
18 - added two regression tests about this in datadesc_usage
19 * Be more verbose when propagating local exceptions
21 * Display the status of simulated processes when receiving SIGINT in
25 * Allow to control the simulation from a trace file.
26 New functions MSG_action_register() and MSG_action_trace_run()
27 The first one allows to associate a function execution to each
28 kind of action while the second one parses a trace file and
29 triggers the corresponding actions within the system.
30 For now, only a toy example is provided in examples/msg/actions
31 * Add an exemple of process migration in examples/msg/migration
32 * Fix a bug in task exchange which broke MSG_task_get_sender()
33 Add a teshsuite regression test for that.
34 [Bug: if MSG_task_get_sender() is called after sender exit,
36 * Fix a bug which prevented suspend/resume to work properly
37 * Display the status of simulated processes when receiving SIGINT
38 This fixes a regression of v3.3. due to the introduction of SIMIX
39 * Bug fixing in failure management:
40 - trace could not start by a failure at time 0
41 - failure during communications were not working
44 * Add SIMIX_process_set_name() to change the name of the current
45 process in the log messages.
48 * Also include strbuff from xbt.h public header
49 * xbt_ex_display(): do not free the exception after displaying
50 This allows to do more with the given exception afterward.
51 Users should call xbt_ex_free() themselves.
54 * Do not require doxygen in maintainer mode
56 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr>
58 SimGrid (3.3) stable; urgency=high
62 * JAVA BINDINGS for MSG (you dreamt of them? We made them)
65 * Introduce the SIMIX module: factorize code between MSG and GRAS.
68 Until now, GRAS were using MSG as an interface to SURF. It was
69 quite difficult because both interface have several differences
70 (MSG channels vs GRAS sockets were the most notable point).
72 This also opens the gate to SMPI (which should occur soon) and speed
73 up simulations by to 40% (even if it were not the main goal).
75 **************************************
76 *DO NOT MIX 3.2 RESULTS WITH 3.3 ONES* Simix may changes simulations!
77 **************************************
78 The point is that events occuring at the exact same timestamp are
79 not scheduled in the same order with the old and new version. This
80 may be enough to completely change the execution of simulations in
81 some cases. Sorry for the inconvenience.
83 * Cleanup and upgrade the XML format to push further scalability
84 issues (check http://hal.inria.fr/inria-00256883/ for more info)
86 * Improve the testing infrastructure with tesh. Now a very large part of
87 the code is tested not only by being run but also by checking that the
88 output match an expected output [Mt].
90 * Move on to FleXML v1.7 for the embeeded XML parsers. This version
91 is really less memory-demanding, which should allow you to use
92 larger files in SimGrid [AL].
94 * Inform valgrind about our contextes, so that it becomes usable
95 with the default (and more effecient) version of SimGrid
96 [contributed by Sékou Diakite, many thanks]
99 * Introduce a listener thread in charge of receiving incomming
100 messages from the network. It allows to overlap communication and
101 computation but most notably, it removes some stupid deadlocks due
102 to the fact that so far, a process could not send and receive at
103 the same time. This made most non trivial communication schema
105 * Convert the PIDs from long int to int to match the MSG ones (and
107 * New function: gras_agent_spawn() to launch a new process on
108 current host. Only working in simulation for now. [Mt]
109 * New function: gras_os_hostport() returning a constant form (ie,
110 not needing to be freed) of "gras_os_hostname():gras_os_myport()"
113 * Make the backtrace of exceptions more human readable [Mt]
114 * New module: xbt/str [Mt]
115 a ton of string utility functions (split, join, printf to a newly
116 allocated buffer, trim, etc)
117 * New module: xbt/hash [Mt]
118 SHA1 hashing algorithm (more to come if needed)
119 * New module: xbt/synchro [Mt]
120 synchronization tools (mutex and conditions) working the same way
121 in simulation and in real life (mainly useful for GRAS, but not
123 * New module: xbt/queue [Mt]
124 classical producer/consumer synchronization scheme
125 * xbt_dynar_new_sync() creates a synchronized dynar. All access
126 (using the classical functions will get serialized) [Mt]
127 * Make dictionary internal table dynamic. No need to specify its size
128 anymore; functions xbt_dict_new_ext() and xbt_dict_hashsize_set()
130 * Make sure the log channels are organized as a tree under windows
131 (because of ANSI C compatibility issue, any channel were child of
135 * Cleaned many thing in surf and fixed a few bugs [AL].
136 * Add a nice command line configuration mechanism to compose models [AL].
137 * Add a new model for parallel tasks (ptask_L07) that is less buggy than
138 the previous one (KCCFLN05). It relies on something that looks like
139 a max-min sharing mechanism but cannot be written as such. A new solver
140 was thus designed [AL].
141 * Add a new solver to lmm. Based on Lagrange optimization and
142 gradient-based descent, it enables to efficiently maximise systems s.a
144 sum f_i(x_i) s.t Ax<= b with A_{i,j}>=0 and f_i a concave function.
146 This solver enables to propose two new network models for TCP Reno and
147 TCP Vegas based on Low's work. These models still need to be fully
148 tested though [Pedro Velho].
151 * Bug fix in SD_simulate. Now the time bound given as argument is
153 * Use the new parallel task model (ptask_L07) as default.
154 * Use the SURF command line configuration mechanism.
155 * 0-size tasks (for synchronization) should now work.
157 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Sun Apr 12 05:20:36 CEST 2009
159 SimGrid (3.2) stable; urgency=high
163 We still experience issues on this platform, but we believe that at
166 GRAS API BREAKAGE (for simplification purpose, sorry):
167 * the gras_msgtype_by_name is not used anymore. Instead of
168 gras_msg_send(toserver, gras_msgtype_by_name("request"), &request);
169 you can write (and must)
170 gras_msg_send(toserver, "request", &request);
171 - If you still want to pass a gras_msgtype_t to the function (to cache
172 the type and avoid the lookup time), use the gras_msg_send_() variant.
173 - Impacted functions:
174 gras_cb_register, gras_cb_unregister, gras_msg_send, gras_msg_wait,
175 gras_msg_rpccall, gras_msg_rpc_async_call, gras_msg_wait_ext
176 * The callbacks are now expected to return 0 when everything went well
177 (just like the main() function)
179 GRAS new features and improvements:
180 * New module mecanism where user code can use per process globals [Mt]
181 This is similar to gras_userdata_*() functions, but for libraries. It
182 factorize some code developped over and over in the examples and AMOK.
183 It has still to be documented and used (only amok/peermanagement is
185 * Fix a vicious bug in the TCP buffering mecanism which leaded to message
186 loss when they were small enough to fit into the buffer and sent quickly
187 enough so that they can all get received in one shoot.
188 * gras_datadesc_by_name and gras_msgtype_by_name: now raise an exception
189 if not found. Use the *_or_null() variant for the old semantic.
190 * In gras_msg_handle, do not discard messages without callback.
191 They are probably messages to be explicitly awaited later (ie, proofs of
192 mis-synchronization in userland since they are sent before being awaited)
194 * gras_socket_meas_send/recv: semantic changed!
195 The numerical arguments used to be (1) the total amount of data to send
196 and (2) msg_size. This was changed to (1) msg_size and (2) amount of
197 messages. This was need for the fool willing to send more than MAXINT
198 bytes on quite fat pipes.
201 * Do really rename the hostmanagement module to peermanagement. [Mt]
202 Ie, rename functions from amok_hm_* to amok_pm_*. This breaks the API,
203 but this is rather new and this was documented in the module
204 documentation (poor excuses, I admit)
205 * Bandwidth measurement semantic changed! This follows the changes to
206 gras_socket_meas_send/recv explained above.
209 * A sequential mode has been added to the workstations. When a workstation
210 is in sequential mode, it can execute only one task, and the other tasks
211 are waiting in a FIFO. [Christophe Thiery]
214 * The KCCFLN05 workstation model now handles parallel tasks. It is the
215 model for SIMDAG. [Christophe Thiery]
216 * Bug fix in the maxmin solver: Some values were close to 0 instead of
217 equal to 0, which caused some bad behaviors in
218 saturated_constraint_set_update. I now use a threshold mechanism like in
222 * When running manually src/testall, you select specific units [Mt]
223 testall is the result of our cunit mecanism, and should replace all
224 the scripty thingy around since bash don't run easily on billware.
226 * A mallocator system has been added. [Christophe Thiery]
227 Mallocators allow you to recycle your unused objects instead of freeing them
228 and allocating new ones.
230 Documentation update:
231 * FAQ reworking + New FAQs:
232 - "Valgrind spits tons of errors!" [Mt]
233 - "How to repport bugs" [Mt]
234 - "Cross-compiling a Windows DLL of SimGrid from Linux" [Mt]
235 - "What is the difference between MSG, SimDag, and GRAS?" [Mt]
236 - Communication time measurement within MSG [AL]
237 - I experience weird communication times when I change the latency [AL]
240 - an introduction to the framework and to the used communication model
241 - an initiatic tour introducing the most proheminent features:
243 . Lesson 0: Installing GRAS
244 . Lesson 1: Setting up your own project
245 o Part 2: Message passing
246 . Lesson 2: Exchanging simple messages
247 . Lesson 3: Passing arguments to the processes (in SG)
248 . Lesson 4: Attaching callbacks to messages
249 . Lesson 5: Using globals in processes
250 . Lesson 6: Logging informations properly
251 . Lesson 7: Using internal timers
252 . Lesson 8: Handling errors through exceptions
253 . Lesson 9: Exchanging simple data
254 . Lesson 10: Remote Procedure Calling (RPC)
255 . Lesson 11: Explicitely waiting for messages
256 . Recapping of message passing features in GRAS
257 - A HOWTO section containing:
258 o HOWTO design a GRAS application
259 More are due, of course. They will come latter. In the meanwhile, you can
260 check the examples which are still here.
262 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri Mar 16 21:11:46 CET 2007
264 SimGrid (3.1) stable; urgency=high
268 There was a stack corruption somewhere, visible only when optimizing
269 with these versions. [Vince]
272 * This is a NEW module! SimDAG (SD for short) is a revival of the old SG
273 module that enabled to play with Directed Acyclic Graphs. It is built
274 directly on top of SURF and provides an API rather close to the old
275 SG. Some old codes using SG are currently under rewrite to check that
276 all needful functions are provided. [Christophe Thiery]
279 * Complete rewrite of the KCCFLN05 workstation model. It is now an
280 extension of the classical CLM03 model that gracefully handles
281 failures. This is now the default model for MSG and GRAS. It doesn't
282 handle parallel tasks yet though. [AL]
283 * Bug fix: Weights were not correctly set in the network part.
284 WARNING: This may have resulted in incorrect results with simulations
285 where there are more than one flow on a given link. [AL]
288 * After a (long ?) discussion on simgrid-devel, we have decided that the
289 convention we had on units was stupid. That is why it has been decided
290 to move from (MBits, MFlops, seconds) to (Bits, Flops, seconds).
291 WARNING : This means that all previous platform files will not work as
292 such with this version! A warning is issued to ask users to update
294 A conversion script can be found in the contrib module of the CVS, under
295 the name contrib/platform_generation/surfxml_update.pl [MQ]
298 * Bug fix: Processes were started in reverse order, wrt deployment file.
299 WARNING: if your code relies on this bug, please fix it. [AL]
300 * Bug fix: Add a test in MSG_task_execute to stop whenever a task is
301 being executed on two different locations. [AL]
302 * Bug fix: Failures are now better supported thanks to Derrick's tests
303 (there was many failure situations I hadn't thought of and that weren't
304 correctly handled). [AL]
305 * New function: MSG_host_is_avail indicates you whether a given m_host_t
309 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
310 Exception occurring on server-side are propagated back to client (!).
312 API CHANGE: the callback changed their prototype. Change:
313 int my_handler(gras_socket_t expeditor, void *payload_data) {
315 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
316 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
318 * New! function: gras_msg_handleall to deal with all messages arriving
319 within a given period.
320 * New! function: gras_socket_server_range to get a server socket in a
321 range of port numbers (ease to avoid port number conflicts) [MQ]
322 * New! gras processes display their backtrace when they get a SIGUSR1
323 or when Ctrl-C is pressed. Use Ctrl-C Ctrl-C to exit.
324 Sweet to debug RL processes [MQ]
328 - Do not force experiment sizes to be expressed in kb, or it becomes
329 impossible to measure the latency this way (needs one byte-long tests)
330 WARNING: this changes the amok_bw_* function semantic. [MQ]
331 - Implements the link saturation stuff. [MQ]
332 * Peer management module:
333 New! module factorizing code that we wrote over and over [MQ].
336 * New module: cunit (my jUnit implementation in ansi C) [MQ]
337 - Test units are placed directly into the library code, they get extracted
338 automatically and placed into the src/testall binary.
339 - Convert most of the XBT tests to this system.
340 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
341 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
342 live in the same C code. This required to change a little bit the API
343 of surfxml parsing but shouldn't be an issue for end-users. [AL]
344 * New module: sparse graph structure with basic algorithms (this is work
345 in progress and the API is not considered to be frozen yet). [AL]
346 * Display more information on backtraces: source line & function names are
347 now displayed just like valgrind does (rely on addr2line tool) [MQ]
348 * New function: xbt_backtrace_display(). Sweet while debuging [MQ]
349 * Reworked a little bit some #include statements to load only required
350 headers. Some user code that relied on SimGrid to include stdlib or
351 stdio may need to include it by themselves. [AL]
352 * Fixed xbt/log.h. A missing SG_BEGIN_DECL prevented compilation with
354 * Renamed xbt_host_t into xbt_peer_t since it betterly describes what I
355 meant. This breaks the API of AMOK and of xbt/config. Sorry about this,
356 but I guess that almost nobody used those parts. [MQ]
358 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 14 Jul 2006 01:32:27 +0200
360 SimGrid (3.0.1) stable; urgency=low
363 * Unfortunately, I had missed 5 misnamed functions:
364 xbt_fifo_item_t xbt_fifo_newitem(void);
365 void xbt_fifo_freeitem(xbt_fifo_item_t);
366 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
367 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
368 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
369 They're now deprecated. Please use their new versions:
370 xbt_fifo_item_t xbt_fifo_new_item(void);
371 void xbt_fifo_free_item(xbt_fifo_item_t);
372 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
373 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
374 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
376 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
377 * Documentation: xbt_log module unmercifully reworked [MQ]
378 * Bugfix: there was a problem with the ending of contexts with
379 the pthread backend. It caused some weird deadlock or behavior
380 depending on the pthread implementation. [AL]
381 * Bugfix: get the exceptions raised in the simulator repport where
382 and why they come from when they are not catched in time [AL, MQ]
385 * Bugfix: Do repport the error when two non-connected hosts try to
386 exchange data (Thanks to Flavien for stumbling into this one) [AL]
389 * Add additionnal checkings on communications. Assert that two
390 communicating hosts are connected by a set of links... [AL]
393 * Add additionnal checkings on channel values in communication [AL]
394 * New: MSG_task_get_source to see on which host a task was generated [HC]
395 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
396 the number of tasks waiting to be received on channel and sent
398 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
399 waits for the first task coming from a given host.. [AL]
401 GRAS new functionnalities: [MQ]
402 * Enhance the parsing macro to allow the size of multidimentional objects
403 to be given thru annotations.
404 * New example (and documentation): Matrix Multiplication a la RPC
405 (as when I was young!) and fix a bunch of bugs found on the way.
407 GRAS performance improvements: [MQ]
409 * Reduce the amount of cbps creation/destruction by making it static to
410 datadesc_send/recv() and using a (newly created) cbps_reset (based on
413 * Change libdata to a set so that we can search for stuff by ID (and thus
414 reduce the insane amount of dict lookups)
417 * Actually implement gras_datadesc_copy() so that we don't have to mimick
418 RL communication on top of SG since it's so uneffective.
419 It may also be used for inter-thread communication in RL, one day.
420 * Use gras_datadesc_copy() to exchange messages on top of SG
422 - improve message exchange performance on top of SG
423 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
424 * Don't exchange on the network the size of the used part of buffer,
425 instead, specify the possible buffer size to read().
427 - reduces the amount of read/write calls (one pair per exchange)
428 - reduces the amount of exchanged data (the size)
429 - allows to retrieve all arrived data on receiver side, if we don't need
430 it right now (subsequent read will peek the buffer)
431 - allows the receiver to proceed with the begining of the stream before
432 everything is arrived
433 - make it possible to build an iov transport (using readv/writev)
435 - take care of the data with non-stable storage (like stacked data),
437 * If possible, TCP send uses vector I/O (when writev() is here)
438 - Don't use it for receive since we send data sizes and data on the
439 same stream, so we wouldn't be able to chain large amount of chunks
440 before having to flush the stuff to read the size.
441 * Rework the transport plugin mecanism to simplify it and reduce the
442 amount of pointer dereferencement when searching for the right function
445 * I guess that now, we do almost as few system calls as possible while
446 doing as few data copy as possible.
448 To improve it further, we could try to send all the sizes first and then
449 all the data (to use iov on receiving size), but it's only a partial
450 solution: when you have 2 dimensional data, the sizes of the second
451 dimension is data of the first dimension, so you need 3 streams.
453 I'm not sure the potential performance gains justify the coding burden.
455 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
457 SimGrid (3.00) stable; urgency=high
460 * New! Give the possibility to hijack the surf parser and thus bypass
461 MSG_create_environment and MSG_launch_application. Have a look at
462 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
464 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
466 SimGrid (2.96) unstable; urgency=low
471 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
472 This deprecates the xbt_error_t mecanisms.
473 It modifies (simplifies) all XBT and GRAS API.
474 MSG API keeps unchanged (exceptions raised by XBT are catched from
475 within MSG and masked with existing error handling facilities)
478 * New! Add a FATPIPE model. [AL]
479 * New! Add a parallel task model. [AL]
480 * New! Add automatically a loopback interface (in the default
481 network model) if none was precised.
484 * Bugfix: MSG_process_resume now works with the current running process.
486 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
487 * Modification of MSG_task_get_compute_duration. Once a task has been
488 processed, the value returned by this function is now equal to 0. [AL]
489 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
490 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
491 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
492 of a m_task. For now, MSG_task_cancel only works with computation
494 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
495 of the processor (in Mflop/s) regardless of the current load on the
497 * API Change: use proper naming convention for MSG_getClock and
498 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
500 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
501 This function changes the priority of a computation task. This priority
502 doesn't affect the transfer rate. A priority of 2 will make a task
503 receive two times more cpu power than the other ones. This function
504 has been added to suit the needs of Nguyen The Loc and hasn't been that
505 much tested yet. So if it fails, please report it and send me your code.
507 * API Change: removed all functions and types that were marked "deprecated"
508 since many months. Renamed MSG_global_init_args to MSG_global_init.
510 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
512 SimGrid (2.95) unstable; urgency=low
515 * Steal some nice code to GNU pth to fix context detection and usage [AL]
516 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
517 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
520 * Bugfix: Allow absolute paths for platform description files [MQ]
521 * Bugfix: do free the variables after use. Leads to drastic performance
523 * Implement max_duration (ie, timeouts) on resources [AL]
526 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
528 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
529 the needed support (timeouts) [AL]
532 * Implement measurement sockets. You can now get the bandwidth between two
533 hosts thanks to AMOK (see below). [MQ]
534 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
535 dynar over the network (yeah) [MQ]
536 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
537 * simple/static token-ring example. [Alexandre Colucci and MQ]
538 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
542 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
543 type parsing. "short" and "long" modifiers were also ignored; other
544 modifier (such as reference level) are still ignored. [MQ]
545 * Embeed the buffer size within the buffer itself on SG. [MQ]
546 That way, send() are atomic and cannot get intermixed anymore (at least
547 the ones which are less than 100k; bigger messages still have the issue)
548 * Array size pushed by the field, not by the field type (or each
549 and every long int will push stuff to the cbps) [MQ]
550 * use select() to sleep since it allows to portably sleep less than one
553 GRAS (minor cleanups)
554 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
557 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
558 * type category 'ignored' killed as it was never used and were difficult
560 * whether a type can cycle or not is now a flag, leaving room for more
561 flags (as "ignored", if we feel the need one day ;)
562 * Rename raw sockets to measurement sockets since "raw" has another
563 meaning in networking community.
566 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
567 and offers features not belonging to GRAS but that most applications
568 need. One day, it may be a set of plugins loadable at runtime.
569 * New module: bandwidth
570 bandwidth measurement between arbitrary nodes running this module. [MQ]
572 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
574 SimGrid (2.94) unstable; urgency=low
576 The first beta release of SimGrid 3 !
580 * Update the main page and the FAQ. Adding references to gforge.
583 * Add a gras_os_getpid function.
586 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
587 * Extend the logs so that they also print PID, hostname, date, ... if
589 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
590 and kill the old version which were in testsuite/
591 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
595 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
596 for users. Instead, each programming environment (GRAS, MSG,...) use it
597 when needed to provide such a feature to users.
598 Don't shortcut the mecanism or you will also shortcut the virtualization
599 you need on the simulator.
603 * Cleanups in configury with regard to compile optimization/warning flags.
604 Also add -fno-loop-optimize to any powerpc since it's the optimization
605 killing gcc (< 3.4.0).
606 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
607 needed by MSG examples complications
608 * Borrow configury beautifications from PHP
611 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
612 warning (thanks loris for stumbling into this one).
613 * Bugfix: stop loading private headers (gras_config.h) from the public
617 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
618 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
619 RL would imply a DNS resolver, which is *hard* to do in a portable way
620 (and therefore delayed).
621 * Implement a real timer mecanism and use it in timing macros. This allows
622 to avoid rounding errors and get a 0.000005 sec precision in timing
623 macros. While I was at it, various cleanups:
624 - allow to declare more than one timed section per file (fix a stupid bug)
625 - move some private declaration to the right place
626 - merge conditional execution and timing macros into emulation module
627 - document the module
628 - make sure the module cleanups its mess on gras_exit
629 * Documentation improvements:
630 - (new) how to compile applications using GRAS
631 - (new) emulation support (timing macros)
633 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
635 SimGrid (2.93) unstable; urgency=low
637 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
640 - Use Paje properly where used. Still to be sanitized properly.
641 - Portability fix: Add an implementation of the contexts using pthread
645 - Add xbt_procname(): returns the name of the current process.
646 Use it to show the current process's name in all logging.
648 - fix detection of older flex version and the reaction, since we do
649 depend on modern ones (we use lex_destroy)
650 - Better separation of SG and RL in the libs: remove all simulation code
651 from libgras. As a result, this lib is now only 200k when stripped.
652 Some of the xbt modules may also be duplicated (two sets and such) and
653 should be cleaned/killed before SG3.
654 - Insist on using xlC on AIX because of weird problems involving gcc there.
655 - Cleanup the make remote stuff. This is now done by scripts
656 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
657 mainly for our private use, but we're working on changing them to user
660 - Bugfix: flush the socket on close only if there is some *output*.
661 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
663 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
666 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
668 SimGrid (2.92) unstable; urgency=low
670 Alpha 3 on the path to SimGrid 3
674 - New! First try of benchmarking macros.
675 - New! First try so that gras_stub_generator generate deployment and
676 remote compilation helpers.
678 - Bugfix: Initialization fix in msg_test.
682 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
684 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
686 - API Change: gras_os_sleep to take the amount of seconds as a double.
687 Accepting an int was error prone since it was the only location where
688 seconds were coded as such. It leaded to damn rounding errors.
689 - Bugfix: Hard to belive that timers ever worked before this.
691 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
693 SimGrid (2.91) unstable; urgency=low
695 Alpha 2 on the path to SimGrid 3
699 - Bug fix in the lmm_solver.
701 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
702 through the function MSG_paje_output.
703 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
704 - It is possible to bound the rate of a communication in MSG with
705 MSG_task_put_bounded() (was already in the previous version but I had forgotten
706 to write it in the changelog).
707 - Bug fix to let GRAS run on top of MSG until we move it directly on top
712 - Various cleanups to the autotools stuff
713 - Begin to move Gras examples to examples/gras/
714 - Let make distcheck work again (yeah!)
716 - documentation overhauled using doxygen.
717 gtk-doc-tools is dead in SimGrid now.
718 - Automatically extract all existing logging categories, and add the list
719 to the documentation (long standing one, to say the less)
721 - Cleanup the known architecture table. Reorder the entries to group what
722 should be, and use a more consistent naming scheme.
723 (some of the test dataset are still to be regenerated)
724 - New! Allow library to register globals on each process just as userdata
726 This is implemented using a xbt_dict and not a xbt_set, so we loose the
727 lookup time (for now).
728 Use it in msg and trp.
729 This cleans a lot the internals and helps enforcing privacy of the
730 headers between the gras components.
731 - New! Add a timer mechanism, not unlike cron(8) and at(1).
732 - Bugfix: gras_os_time was delirious in RL.
733 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
735 - Reenable GRAS now that it works.
737 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
739 SimGrid (2.90) unstable; urgency=low
741 Alpha 1 on the path to SimGrid 3
743 * It is a long time since the last release of SimGrid. I'm sorry about
744 that but as I had told you, I was rewriting a lot of things. I apologize
745 to those who had been reporting bugs to me and that I had not answered.
746 If your bug is still in the new version, please tell me. Here is a
747 summary of the main changes.
749 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
750 lead by Martin Quinson. As a consequence SimGrid gains a lot in
751 portability, speed, and a lot more but you'll figure it out later.
752 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
753 I wanted to release the new MSG as soon as possible and I have
754 broken GRAS, which is the reason why, for now, only MSG is fully
755 functional. A laconic description of these projects is available
756 in the documentation.
758 * REVOLUTION 2: I have removed SG and I am now using a new simulation
759 kernel optimized for our needs (called SURF but only the developers
760 should use it). Hence, MSG is now roughly 30 times faster and I think
761 that by rewriting a little bit MSG, I could event speed it up a little
762 bit more. Beside the gain in speed, it is also much easier to encode a
763 new platform model with SURF than it was with SG. More to come...
765 * REVOLUTION 3: I have tried to change a little as possible the API of
766 MSG but a few things really had to disappear. The main differences
767 with the previous version are :
768 1) no more m_links_t and the corresponding functions. Platforms are
769 directly read from a XML description and cannot be hard-coded
770 anymore. The same format is used for application deployment
771 description. The new format is described in the documentation.
772 Have a look in tools/platform_generation. There is a tiny script
773 that converts from the old platform format to the new one. Concerning
774 the application deployment format, parsing the old one is tricky.
775 I think most of you should however be able to convert your files. If
776 it is really an issue, I can write a C code that does the conversion.
778 2) the toolbox tbx does not exist anymore. We now have a library
779 with much more data-structures but without the hash-tables (we have
780 dictionaries that are much faster).
782 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
784 *****************************************************************************
785 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
786 * don't feel like converting the oldies. (Mt) *
787 *****************************************************************************
790 Version 2.90: "the long awaited one"
791 - Finished rewriting and debugging MSG. Rewrote the documentation.
792 - disable GRAS for now since it needs to be ported to the newest SG
795 - Finish the port to windows (using mingw32 for cross-compile)
798 - Main loop and datastructures of SURF. A cpu resource object is
799 functional. Surf can thus be used to create cpu's with variable
800 performance on which you can execute some actions.
802 2004-11-15 Martin Quinson
803 - Port to ARM. Simply added the alignment and size descriptions. Should
804 work, but the ARM machines are so slow that I didn't had the opportunity
805 to 'make check' over there yet.
807 2004-11-15 Arnaud Legrand
808 - Trace manager now written. It uses a heap structure and is therefore
809 expected to be efficient. It may however be speeded up (particularly
810 when many events occur at the same date) by using red and black
811 trees. One day maybe...
812 - Max-min linear system solver written. It uses a sparse matrix
813 structure taking advantage of its expected use. Most operations are
814 O(1) and free/calloc are called as few as possible. The computation of
815 the minimum could however be improved by using a red and black tree
818 2004-11-03 Arnaud Legrand
819 - Rename every gras_* function that was in xbt/ to its xbt_
821 - Add a heap and a doubly-linked list to xbt
822 - Added a dichotomy to the dictionaries. make check works as well before
823 so I assume that the patch is correct. I do not know however if things
824 run effectively faster than before now. :)
826 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
827 SimGrid, and the version number is bumped to 2.x
829 2004-10-29 Martin Quinson
830 - Introduction of the remote errors.
831 They are the result of a RMI/RPC on the remote machine.
832 ErrCodes being scalar values, you can't get the host on which those
833 errors did happen. Extending the error mechanism as in Gnome is possible.
834 No idea yet whether it is a good idea.
836 2004-10-28 Martin Quinson
837 - Interface revolution: the Starred Structure Eradication.
838 I used to do typedef struct {} toto_t; and then handle *toto_t.
839 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
840 - ???_t is a valid type (builded with typedef)
841 - s_toto_t is a structure (access to fields with .)
842 - s_toto is a structure needing 'struct' keyword to be used
843 - e_toto_t is an enum
844 - toto_t is an 'object' (struct*)
846 typedef struct s_toto {} s_toto_t, *toto_t;
847 typedef enum {} e_toto_t;
848 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
849 s_toto_t) is private.
851 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
852 so that it can set the variable to NULL. It was so for dicts and sets,
853 it changed for dynars.
855 - Fix a bunch of memleaks in dict_remove
856 - Fix a bug in sg/server_socket opening: it failed all the time.
858 2004-10-07 Martin Quinson
859 - Speed up dynar lookup operation a bit.
861 gras_dynar_get is dead.
863 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
864 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
865 (which gives you the address of the stored data).
867 gras_dynar_get_as is an helpful macro which allows you to retrieve a
868 copy of the data using an affectation to do the job and not a memcpy.
870 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
871 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
873 It does not really speedup the dynar test because they are
874 setting elements all the time (and look them seldom). But the dict does
875 far more lookup than setting.
877 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
879 2004-10-05 Martin Quinson
880 - Allow to (en/dis)able the cycle detection at run time.
882 Whether we should check for cycle or not is now a property of each
883 datatype. When you think there may be some cycle, use datadesc_cycle_set.
884 datadesc_cycle_unset allow to remove this property when previously set.
886 Note that the cycle detection is off by default since it impacts the
887 performance. Watch the data you feed GRAS with ;)
889 This property is hereditary. Any element embedded in a structure having it
890 set have it set for the time of this data exchange.
892 You should set it both on sender and receiver side. If you don't set it on
893 sender side, it will enter an endless loop. If you forget on receiver
894 side, the cycles won't be recreated after communication.
896 - Header reorganization.
897 Kill gras_private.h, each submodule must load the headers it needs.
899 2004-10-04 Martin Quinson
900 - Interface revolution: do not try to survive to malloc failure.
902 Now, gras_malloc and friends call gras_abort() on failure.
903 As a conclusion, malloc_error is not a valid error anymore, and all
904 functions for which it was the only gras_error_t return value are
905 changed. They now return void, or there result directly.
906 This simplify the API a lot.
908 2004-09-29 Martin Quinson
909 - Re-enable raw sockets.
910 Created by gras_socket_{client,server}_ext;
911 Used with gras_raw_{send,recv}
914 It should allow to kill the last bits of gras first version soon.
916 This is not completely satisfactory yet (duplicate code with
917 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
920 - Simplify transport plugin (internal) interface by not passing any
921 argument to _server and _client, but embedding them in the socket
924 2004-09-28 Martin Quinson
925 - Finish the port to AIX.
926 autoconf was my problem (segfault within the malloc replacement
927 function. No idea why)
929 2004-09-16 Martin Quinson
930 - Fix some size_t madness on 64bit architectures.
932 2004-09-08 Martin Quinson
933 - Reduce the number of system headers loaded, overload some more system
934 calls (such as malloc to cast the result of the system one, and work
936 - Fix and reintroduce the config support
938 2004-09-07 Martin Quinson
939 - Source code reorganization to allow Arnaud to surf all over there.
940 - Allow to document the logging categories.
941 - Remove all uppercase from logging categories and useless cleanup in names.
943 2004-08-18 Martin Quinson
944 Version 0.6.2 (protocol not changed; API changed)
945 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
946 gras_error_t), and NULL when not found. Functions expecting a msgtype
947 as argument (msg_wait; msg_send) deal with NULL argument by providing a
948 hopefully usefull message.
949 - Portability to prehistoric sparcs again
951 2004-08-17 Martin Quinson
952 Version 0.6.1 (protocol not changed; ABI not changed)
953 - prealloc some buffers to speed things up
955 2004-08-11 Martin Quinson
956 Version 0.6 (protocol not changed; ABI expended)
957 - The parsing macro can deal with the references, provided that you add
958 the relevant annotations (using GRAS_ANNOTE(size,field_name))
960 2004-08-09 Martin Quinson
961 Version 0.5 (protocol not changed; ABI changed)
962 - Allow to off turn the cycle detection code in data exchange at
963 compilation time. It should be at run time, but I'm short of time (and
964 the config stuff is still broken). That way, we keep dict out of the
965 critical path, which is good because the performance is poor:
966 - search not dichotomial yet
967 - dynar give no way to access their content and memcpy everytime
968 - In composed data description (struct, ref and so on), stop foolness of
969 keeping the subtype's ID, but store the type itself. This keeps sets out
970 of the critical path, which is good since they rely on dynar and
971 dictionnaries. The only loose of that is that we cannot detect the
972 redeclaration of a structure/union with another content (but I'm not sure
973 the code detected well this error before anyway). We still can detect
974 the redefinition discrepancy for the other types.
975 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
976 it breaks the code because of type-punning used all over the place).
977 This breaks on all non-gcc architectures (for now).
979 All those changes (plus the buffer of last time) allow me to gain 2 order
980 of magnitude on cruel tests consisting of 800000 array of integers on two
981 level of a hierarchical structure (200 secondes -> 4 secondes)
984 - the selector of reference must now return the type it points to, not
987 2004-08-06 Martin Quinson
988 Version 0.4 (protocol changed; ABI not changed)
989 - Allow to pass --gras-log argument to processes in simulation mode. Really.
990 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
991 - Implement a buffer transport, and use it by default (it relies on tcp in
992 real life and on sg in simulation).
993 That's a bit hackish since I had a new field to the structure to store
994 its data without interfering with the subtype ones. Inheritance
995 is tricky in C. And that's a kind of reverse inheritance with one class
996 derivating two classes. Or maybe a game with java interfaces. Anyway,
997 that's damn hard in C (at least).
998 Moreover, I got tired while trying to ensure plugin separation and
999 genericity in SG mode. MSG wants me to do weird things, so let's go for
1000 cruel hacks (temporarily of course ;).
1001 See comment in transport_private.h:71
1002 - do not include all the _interface headers in private but in the files
1003 which really need them (to cut the compilation time when they are
1006 2004-07-26 Martin Quinson
1007 Version 0.3 (protocol not changed; ABI changed)
1008 - Major overhault of the datadesc interface to simplify it:
1009 - shorted the function names:
1010 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
1011 - add a trivial way to push/pop integers into the cbps without malloc.
1012 This allows to make really generic sub_type description, which simply
1013 pop their size of the stack.
1014 - add a function gras_datadesc_ref_pop_arr() which does what users want
1015 most of the time: Declare a dynamic array (which pops its size of the
1016 stack) and declare a reference to it. Poor name, but anyway.
1017 - kill the post-send callback, add a post-receive one
1019 2004-07-23 Martin Quinson
1020 Version 0.2 (protocol changed; ABI changed)
1021 - add some testing for cpbs in the test cases, and fix some more bugs.
1022 This invalidate again the little64 data file, since I cannot regenerate
1024 - remove an awfull optimization in the logging stuff, allowing me to:
1025 - understand it again
1026 - learn gcc how to check that the argument match the provided format
1027 - fix all errors revealed by gcc after that
1028 - internal keys of dict are not \0 terminated. Deal with it properly in
1029 loggings instead of segfaulting when the user want to see the logs :-/
1031 2004-07-22 Martin Quinson
1032 - Fix some stupid bug preventing cbps (callback postit) from working
1034 2004-07-21 Martin Quinson
1035 - Some documentation cleanups
1036 - remove the useless last argument of msgtype_declare
1037 - rename the Virtu functions to fit into the 'os' namespace
1038 - move headers src/include -> src/include/gras/ and stop fooling with
1039 gras -> . symbolic link
1040 - make distcheck is now successful
1042 2004-07-19 Martin Quinson
1044 - Build shared library also
1045 - Install html doc to the right location
1046 - stop removing maintainer files in make clean
1047 - build tests only on make check
1049 2004-07-13 Martin Quinson
1051 - No major issue in previous version => change versionning schema
1052 - Re-enable little64 convertion test now that Abdou kindly regenerated the
1053 corresponding dataset.
1055 2004-07-11 Martin Quinson
1057 - Get it working with any kind of structure (we can compute the padding
1058 bytes remotely for all the architectures I have access to)
1059 - Implement the structure parsing macro (still not quite robust/complete)
1060 - Improvement to the remote testing toysuite
1062 2004-07-10 Martin Quinson
1063 [autoconf mechanism]
1064 - get ride of a bunch of deprecated macros
1065 - actually run the test for two-compliment, not only compile it :-/
1066 - test whether the structures get packed (and bail out if yes. Damn.
1067 Alignment is a serious matter)
1068 - test whether the structures get compacted (but respecting the alignment
1069 constraints of each types)
1070 - test whether the array fields of structures can straddle alignment boundaries
1072 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
1073 (took me 2 hours to find that bug, looking at the wrong place)
1074 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
1075 before sending/receiving and are used to compute the offsets of fields
1076 - Given that padding size depend even on compiler options, keep track of
1077 alignment and aligned_size only for the current architecture. Not a big
1078 deal since we send structure fields one after the other (seems
1080 - Add the datastructure used for IEEE paper by the PBIO guys to the test
1081 program, let it work on linux/gcc/little32. portability todo.
1083 2004-07-08 Martin Quinson
1084 - import and improve remote compilation support from FAST
1085 - make sure make check works on half a dozen of machines out there
1087 2004-07-07 Martin Quinson
1088 Let's say it's version 0.0.3 ;)
1089 - Implement conversions (yuhu!)
1090 - Let it work on solaris (beside conversion, of course)
1091 - Stupid me, using rand() to generate the conversion datatests in not wise.
1093 2004-07-06 Martin Quinson
1094 - Let make dist work, since I'm gonna need it to compile on remote hosts
1095 - Let Tests/datadesc_usage write the architecture on which the file was
1096 generated as first byte.
1097 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
1098 architecture descriptions.
1099 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
1100 on those architectures.
1101 - Optimization: send/recv array of scalar in one shoot
1103 2004-07-05 Martin Quinson
1104 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
1106 - Plug a whole bunch of memleaks
1107 - each process now have to call gras_{init,exit}. One day, their log
1108 settings will be separated
1109 - Continue the code factorisation between SG, RL and common in Transport.
1111 2004-07-04 Martin Quinson
1113 - Redistribution between SG and RL.
1114 We wanna have to accept in SG, so move accepted related parts of RL in
1115 the common part. (more precisely, the dynar of all known sockets is no
1116 more a static in transport.c, but part of the process_data)
1118 [gras_stub_generator]
1119 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
1121 2004-07-03 Martin Quinson
1122 - Create a new log channel tbx containing dict, set, log, dynar (to shut
1123 them all up in one shot)
1125 - Fix the ugly case of reference to dynamic array.
1126 - New (semi-public) function gras_datadesc_size to allow the messaging
1127 layer to malloc the needed space for the buffer.
1129 - gras_socket_close now expect the socket to close (and not its address to
1130 put NULL in it after it). This is because the socket passed to handlers
1131 is one of their argument (=> not writable).
1133 - propagate the interface cleanup from last week in datadesc, ie remove a
1134 superfluous level of indirection. User pass adress of variable
1135 containing data (both when sending and receiving), and not of a variable
1136 being a pointer to the data. Let's say that I like it better ;)
1137 The price for that is constructs like "int msg=*(int*)payload" in
1138 handlers, but it's a fine price, IMHO.
1140 - Let it work in RL (yuhu)
1142 2004-06-21 Martin Quinson
1144 - porting SG plugin and SG select to new standards (works almost).
1145 - plug memleaks and fix bugs around.
1148 - cleanup the prototype of data recv and force users to specify when they
1149 want to handle references to objects. Test case working even for cycles.
1150 - plug memleaks. Valgrind is perfectly ok with this.
1152 2004-06-12 Martin Quinson
1154 - cleanup the separation between plugin and main code in plugin creation
1156 2004-06-11 Martin Quinson
1158 - Reput hook for raw sockets, needed for BW experiments
1159 - kill a few lines of dead code
1160 [Data description] Interface cleanup
1161 - gras_datadesc_by_name returns the searched type or NULL.
1162 That way, no variable is needed to use a type desc once, which makes
1164 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
1165 two parameters were strings (field name, type name), leading to
1167 [Dicos] Interface cleanup
1168 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
1169 This is consistant with the dynar API.
1171 2004-04-21 Martin Quinson
1173 - Porting to new standards.
1175 - interface cleanup.
1176 There is no bag anymore, no need to take extra provision to mask the
1177 pointers behind "ID".
1178 Better splitup of functions between files create/exchange/convert.
1179 This is still a bit artificial since convert and receive are so
1180 interleaved, but anyway.
1182 - add a queued message list to procdata (the ones not matching criteria
1184 - factorize some more code between SG and RL wrt procdata
1186 - use gras_exit in example to track memleaks
1187 - get rid of gs_example now that GS is properly integrated into gras
1188 - update run_test to integrate the lastest tests (datadesc)
1190 - rename WARNINGn macros to WARNn since it prooved error-prone
1192 2004-04-19 Martin Quinson
1194 - register init/exit functions within gras module mechanism
1195 - send/receive function.
1196 Convertion is not implemented, but short-cutted if not needed.
1197 struct/array elements are sent one by one (instead of block-wise), but
1198 nobody really cares (yet). Get a prototype before optimizing.
1199 - tests (using a file socket) for DD send/receive on:
1200 - base types: int, float
1201 - array: fixed size, string (ie ref to dynamic string)
1202 - structure: homogeneous, heterogeneous
1203 - chained list, graph with cycle
1204 Believe it or not, valgrind is not too unhappy with the results. The
1205 cycle happily segfaults, but the others are ok. And I'm sick of pointers
1209 - Bugfix when using a filename explicitely (instead of '-')
1211 2004-04-09 Martin Quinson
1213 - factorize more code between RL and SG in socket creation
1214 - Complete the implementation and tests of:
1216 o file (only in RL, and mainly for debugging)
1218 I lost 3 days to design a portable address resolver, and then decided
1219 that the prototype mainly have to run on my box.
1220 Addressing portability too early may be like optimizing too early :-/
1222 - use gras_init in the Tests instead of the crappy parse_log_opt
1223 (the latter function is removed)
1224 [Conditional execution]
1225 - New functions: gras_if_RL/gras_if_SG (basic support for this)
1226 [Code reorganisation]
1227 - Get rid of libgrasutils.a since it makes more trouble than it solves.
1228 Build examples against the RL library, since there is no way to disable
1229 its creation for now.
1231 For information, the beginning of coding on GRAS was back in june
1232 2003. I guess that every line has been rewritten at least twice since