1 SimGrid (3.3.2-svn) unstable; urgency=low
4 * Implement some more MPI primitives:
5 MPI_Waitany, MPI_Waitall, MPI_Reduce, MPI_Allreduce
6 * Add support for optimized collectives (Bcast is now binomial by default)
7 * Port smpirun and smpicc to OS X
10 * Extract the routing logic into its own object.
11 (was dupplicated in network.c and workstation_LV07.c;
12 Allows to implement other ways of storing that info)
13 => kill now useless network_card concept
14 - Use dynar to represent routes (instead of void** + int*)
15 - kill link_set (use surf_network_model->resource_set instead)
16 - Add a command-line option to choose the routing schema to use
18 * Simplify model declaration
19 (less redirections, less function to write when defining a model)
20 - Factorize stuff between models:
23 surf_model_resource_set(model)
24 surf_model_resource_by_name(model, name)
25 - Unify the types of models in s_surf_model_t (using an union)
26 - Embeed fields of common_public directly into s_surf_model_t
27 - Rename model methods:
28 action_free ~> action_unref
29 action_change_state ~> action_state_set
30 action_get_state ~> action_state_get
31 - Change model methods into functions :
32 (model)->common_public->action_use ~> surf_action_ref
34 * Implement a generic resource; use it as ancestor to specific ones
35 (allows to kill duplicated code in models)
36 Drawback: timer command don't need no name nor properties;
37 workstation_CLM03 don't need no properties
38 (but I guess we can live with those few bytes wasted)
40 * Improve the action object model
41 - implement a constructor avoiding dupplicated code about field
42 initialization in generic_action part.
45 * Kill SD_link_get_properties: hard to maintain and makes very little sense
46 Shout out if you used it.
49 * Add xbt_set_get_by_name_or_null() [Silas De Munck]
50 * Add xbt_graph_node_get_outedges() [Silas De Munck]
52 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr>
54 SimGrid (3.3.1) stable; urgency=low
57 * Implement a --cfg-help to show existing configuration variables
58 * Build chain do not require doxygen in maintainer mode
61 * fix a bug on struct sizeof computation, which prevented the
62 exchange of arrays of structs in some conditions
63 - added a regression test about this in datadesc_usage
64 * Allow the exchange of 0-long dynamic vectors.
65 - for that, use -1 as indicator of dynamic size instead of 0
66 - This implied to change any size from unsigned long to long,
67 reducing a bit communication abilities, but I guess that with
68 64bits being quite common, this is more than enough.
69 - This also induce a protocol change, thus bumping network protocol
70 version from 0 to 1 (if we have external users, we have to get
71 clean on that point too ;)
72 - added two regression tests about this in datadesc_usage
73 * Be more verbose when propagating local exceptions
75 * Display the status of simulated processes when receiving SIGINT in
79 * Allow to control the simulation from a trace file.
80 New functions MSG_action_register() and MSG_action_trace_run()
81 The first one allows to associate a function execution to each
82 kind of action while the second one parses a trace file and
83 triggers the corresponding actions within the system.
84 For now, only a toy example is provided in examples/msg/actions
85 * Add an exemple of process migration in examples/msg/migration
86 * Fix a bug in task exchange which broke MSG_task_get_sender()
87 Add a teshsuite regression test for that.
88 [Bug: if MSG_task_get_sender() is called after sender exit,
90 * Fix a bug which prevented suspend/resume to work properly
91 * Display the status of simulated processes when receiving SIGINT
92 This fixes a regression of v3.3. due to the introduction of SIMIX
93 * Bug fixing in failure management:
94 - trace could not start by a failure at time 0
95 - failure during communications were not working
98 * Add SIMIX_process_set_name() to change the name of the current
99 process in the log messages.
100 * Store smx_hosts in a dict since we only retrieve them by name
101 * Move the configuration infrastructure to surf
104 * Move the configuration infrastructure to surf
107 * Massive internal cleanups:
108 - Store internal structures on processes instead of hosts (allows
109 to have more than one process per host, in addition of being more
111 - Cleanup the initialization/finalization process
112 - Kill a whole bunch of unneeded synchronization:
113 processes run in exclusive manner within the simulator
114 - Move queues from global tables to process data fields
116 - now accept -platform and -hostfile arguments
117 - Pass the right rank value to processes according to the hostfile
118 * Compile the examples by default, and use them as regression tests
119 * Implement MPI_Wtime()
120 * Change the reference speed to a command line option
123 * TCP_gamma can now be specified as command line option using
124 --cfg=TCP_gamma:10000000.0
125 * Change the --surf-path cmd line option into --cfg=path:
128 * Also include strbuff from xbt.h public header
129 * xbt_ex_display(): do not free the exception after displaying
130 This allows to do more with the given exception afterward.
131 Users should call xbt_ex_free() themselves.
135 Portability report of this version:
136 * Main portability targets:
137 - Linux(debian)/x86/context
138 - Linux(debian)/x86/pthreads
139 - Linux(debian)/amd64/context
140 - Linux(debian)/amd64/pthreads
141 These targets fail about 1/10 of times on gras/pmm, but we believe
142 that this is because of the test, not because of simgrid.
143 amok/saturate_sg fails even more rarely, and the test may not be
146 - Mac OSX Leopard/x86/context
147 The test suite still spits tons of errors because some obscure
148 force prevents us from removing the temporary directories
149 arguing that they still contain some metadata I've never heard of.
150 Smpi fails because seq is not installed.
151 Everything seems to work properly beside of that.
154 - AIX version 5.3 (both contexts and pthread)
155 Smpi still fails there because mktemp is not installed.
156 XML inclusions seems rosty on AIX.
158 * Windows: it's still lagging behind. If you want to help, please
161 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Sat, 27 Jun 2009 00:14:30 +0200
163 SimGrid (3.3) stable; urgency=high
167 * JAVA BINDINGS for MSG (you dreamt of them? We made them)
170 * Introduce the SIMIX module: factorize code between MSG and GRAS.
173 Until now, GRAS were using MSG as an interface to SURF. It was
174 quite difficult because both interface have several differences
175 (MSG channels vs GRAS sockets were the most notable point).
177 This also opens the gate to SMPI (which should occur soon) and speed
178 up simulations by to 40% (even if it were not the main goal).
180 **************************************
181 *DO NOT MIX 3.2 RESULTS WITH 3.3 ONES* Simix may changes simulations!
182 **************************************
183 The point is that events occuring at the exact same timestamp are
184 not scheduled in the same order with the old and new version. This
185 may be enough to completely change the execution of simulations in
186 some cases. Sorry for the inconvenience.
188 * Cleanup and upgrade the XML format to push further scalability
189 issues (check http://hal.inria.fr/inria-00256883/ for more info)
191 * Improve the testing infrastructure with tesh. Now a very large part of
192 the code is tested not only by being run but also by checking that the
193 output match an expected output [Mt].
195 * Move on to FleXML v1.7 for the embeeded XML parsers. This version
196 is really less memory-demanding, which should allow you to use
197 larger files in SimGrid [AL].
199 * Inform valgrind about our contextes, so that it becomes usable
200 with the default (and more effecient) version of SimGrid
201 [contributed by Sékou Diakite, many thanks]
204 * Introduce a listener thread in charge of receiving incomming
205 messages from the network. It allows to overlap communication and
206 computation but most notably, it removes some stupid deadlocks due
207 to the fact that so far, a process could not send and receive at
208 the same time. This made most non trivial communication schema
210 * Convert the PIDs from long int to int to match the MSG ones (and
212 * New function: gras_agent_spawn() to launch a new process on
213 current host. Only working in simulation for now. [Mt]
214 * New function: gras_os_hostport() returning a constant form (ie,
215 not needing to be freed) of "gras_os_hostname():gras_os_myport()"
218 * Make the backtrace of exceptions more human readable [Mt]
219 * New module: xbt/str [Mt]
220 a ton of string utility functions (split, join, printf to a newly
221 allocated buffer, trim, etc)
222 * New module: xbt/hash [Mt]
223 SHA1 hashing algorithm (more to come if needed)
224 * New module: xbt/synchro [Mt]
225 synchronization tools (mutex and conditions) working the same way
226 in simulation and in real life (mainly useful for GRAS, but not
228 * New module: xbt/queue [Mt]
229 classical producer/consumer synchronization scheme
230 * xbt_dynar_new_sync() creates a synchronized dynar. All access
231 (using the classical functions will get serialized) [Mt]
232 * Make dictionary internal table dynamic. No need to specify its size
233 anymore; functions xbt_dict_new_ext() and xbt_dict_hashsize_set()
235 * Make sure the log channels are organized as a tree under windows
236 (because of ANSI C compatibility issue, any channel were child of
240 * Cleaned many thing in surf and fixed a few bugs [AL].
241 * Add a nice command line configuration mechanism to compose models [AL].
242 * Add a new model for parallel tasks (ptask_L07) that is less buggy than
243 the previous one (KCCFLN05). It relies on something that looks like
244 a max-min sharing mechanism but cannot be written as such. A new solver
245 was thus designed [AL].
246 * Add a new solver to lmm. Based on Lagrange optimization and
247 gradient-based descent, it enables to efficiently maximise systems s.a
249 sum f_i(x_i) s.t Ax<= b with A_{i,j}>=0 and f_i a concave function.
251 This solver enables to propose two new network models for TCP Reno and
252 TCP Vegas based on Low's work. These models still need to be fully
253 tested though [Pedro Velho].
256 * Bug fix in SD_simulate. Now the time bound given as argument is
258 * Use the new parallel task model (ptask_L07) as default.
259 * Use the SURF command line configuration mechanism.
260 * 0-size tasks (for synchronization) should now work.
262 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Sun Apr 12 05:20:36 CEST 2009
264 SimGrid (3.2) stable; urgency=high
268 We still experience issues on this platform, but we believe that at
271 GRAS API BREAKAGE (for simplification purpose, sorry):
272 * the gras_msgtype_by_name is not used anymore. Instead of
273 gras_msg_send(toserver, gras_msgtype_by_name("request"), &request);
274 you can write (and must)
275 gras_msg_send(toserver, "request", &request);
276 - If you still want to pass a gras_msgtype_t to the function (to cache
277 the type and avoid the lookup time), use the gras_msg_send_() variant.
278 - Impacted functions:
279 gras_cb_register, gras_cb_unregister, gras_msg_send, gras_msg_wait,
280 gras_msg_rpccall, gras_msg_rpc_async_call, gras_msg_wait_ext
281 * The callbacks are now expected to return 0 when everything went well
282 (just like the main() function)
284 GRAS new features and improvements:
285 * New module mecanism where user code can use per process globals [Mt]
286 This is similar to gras_userdata_*() functions, but for libraries. It
287 factorize some code developped over and over in the examples and AMOK.
288 It has still to be documented and used (only amok/peermanagement is
290 * Fix a vicious bug in the TCP buffering mecanism which leaded to message
291 loss when they were small enough to fit into the buffer and sent quickly
292 enough so that they can all get received in one shoot.
293 * gras_datadesc_by_name and gras_msgtype_by_name: now raise an exception
294 if not found. Use the *_or_null() variant for the old semantic.
295 * In gras_msg_handle, do not discard messages without callback.
296 They are probably messages to be explicitly awaited later (ie, proofs of
297 mis-synchronization in userland since they are sent before being awaited)
299 * gras_socket_meas_send/recv: semantic changed!
300 The numerical arguments used to be (1) the total amount of data to send
301 and (2) msg_size. This was changed to (1) msg_size and (2) amount of
302 messages. This was need for the fool willing to send more than MAXINT
303 bytes on quite fat pipes.
306 * Do really rename the hostmanagement module to peermanagement. [Mt]
307 Ie, rename functions from amok_hm_* to amok_pm_*. This breaks the API,
308 but this is rather new and this was documented in the module
309 documentation (poor excuses, I admit)
310 * Bandwidth measurement semantic changed! This follows the changes to
311 gras_socket_meas_send/recv explained above.
314 * A sequential mode has been added to the workstations. When a workstation
315 is in sequential mode, it can execute only one task, and the other tasks
316 are waiting in a FIFO. [Christophe Thiery]
319 * The KCCFLN05 workstation model now handles parallel tasks. It is the
320 model for SIMDAG. [Christophe Thiery]
321 * Bug fix in the maxmin solver: Some values were close to 0 instead of
322 equal to 0, which caused some bad behaviors in
323 saturated_constraint_set_update. I now use a threshold mechanism like in
327 * When running manually src/testall, you select specific units [Mt]
328 testall is the result of our cunit mecanism, and should replace all
329 the scripty thingy around since bash don't run easily on billware.
331 * A mallocator system has been added. [Christophe Thiery]
332 Mallocators allow you to recycle your unused objects instead of freeing them
333 and allocating new ones.
335 Documentation update:
336 * FAQ reworking + New FAQs:
337 - "Valgrind spits tons of errors!" [Mt]
338 - "How to repport bugs" [Mt]
339 - "Cross-compiling a Windows DLL of SimGrid from Linux" [Mt]
340 - "What is the difference between MSG, SimDag, and GRAS?" [Mt]
341 - Communication time measurement within MSG [AL]
342 - I experience weird communication times when I change the latency [AL]
345 - an introduction to the framework and to the used communication model
346 - an initiatic tour introducing the most proheminent features:
348 . Lesson 0: Installing GRAS
349 . Lesson 1: Setting up your own project
350 o Part 2: Message passing
351 . Lesson 2: Exchanging simple messages
352 . Lesson 3: Passing arguments to the processes (in SG)
353 . Lesson 4: Attaching callbacks to messages
354 . Lesson 5: Using globals in processes
355 . Lesson 6: Logging informations properly
356 . Lesson 7: Using internal timers
357 . Lesson 8: Handling errors through exceptions
358 . Lesson 9: Exchanging simple data
359 . Lesson 10: Remote Procedure Calling (RPC)
360 . Lesson 11: Explicitely waiting for messages
361 . Recapping of message passing features in GRAS
362 - A HOWTO section containing:
363 o HOWTO design a GRAS application
364 More are due, of course. They will come latter. In the meanwhile, you can
365 check the examples which are still here.
367 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri Mar 16 21:11:46 CET 2007
369 SimGrid (3.1) stable; urgency=high
373 There was a stack corruption somewhere, visible only when optimizing
374 with these versions. [Vince]
377 * This is a NEW module! SimDAG (SD for short) is a revival of the old SG
378 module that enabled to play with Directed Acyclic Graphs. It is built
379 directly on top of SURF and provides an API rather close to the old
380 SG. Some old codes using SG are currently under rewrite to check that
381 all needful functions are provided. [Christophe Thiery]
384 * Complete rewrite of the KCCFLN05 workstation model. It is now an
385 extension of the classical CLM03 model that gracefully handles
386 failures. This is now the default model for MSG and GRAS. It doesn't
387 handle parallel tasks yet though. [AL]
388 * Bug fix: Weights were not correctly set in the network part.
389 WARNING: This may have resulted in incorrect results with simulations
390 where there are more than one flow on a given link. [AL]
393 * After a (long ?) discussion on simgrid-devel, we have decided that the
394 convention we had on units was stupid. That is why it has been decided
395 to move from (MBits, MFlops, seconds) to (Bits, Flops, seconds).
396 WARNING : This means that all previous platform files will not work as
397 such with this version! A warning is issued to ask users to update
399 A conversion script can be found in the contrib module of the CVS, under
400 the name contrib/platform_generation/surfxml_update.pl [MQ]
403 * Bug fix: Processes were started in reverse order, wrt deployment file.
404 WARNING: if your code relies on this bug, please fix it. [AL]
405 * Bug fix: Add a test in MSG_task_execute to stop whenever a task is
406 being executed on two different locations. [AL]
407 * Bug fix: Failures are now better supported thanks to Derrick's tests
408 (there was many failure situations I hadn't thought of and that weren't
409 correctly handled). [AL]
410 * New function: MSG_host_is_avail indicates you whether a given m_host_t
414 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
415 Exception occurring on server-side are propagated back to client (!).
417 API CHANGE: the callback changed their prototype. Change:
418 int my_handler(gras_socket_t expeditor, void *payload_data) {
420 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
421 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
423 * New! function: gras_msg_handleall to deal with all messages arriving
424 within a given period.
425 * New! function: gras_socket_server_range to get a server socket in a
426 range of port numbers (ease to avoid port number conflicts) [MQ]
427 * New! gras processes display their backtrace when they get a SIGUSR1
428 or when Ctrl-C is pressed. Use Ctrl-C Ctrl-C to exit.
429 Sweet to debug RL processes [MQ]
433 - Do not force experiment sizes to be expressed in kb, or it becomes
434 impossible to measure the latency this way (needs one byte-long tests)
435 WARNING: this changes the amok_bw_* function semantic. [MQ]
436 - Implements the link saturation stuff. [MQ]
437 * Peer management module:
438 New! module factorizing code that we wrote over and over [MQ].
441 * New module: cunit (my jUnit implementation in ansi C) [MQ]
442 - Test units are placed directly into the library code, they get extracted
443 automatically and placed into the src/testall binary.
444 - Convert most of the XBT tests to this system.
445 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
446 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
447 live in the same C code. This required to change a little bit the API
448 of surfxml parsing but shouldn't be an issue for end-users. [AL]
449 * New module: sparse graph structure with basic algorithms (this is work
450 in progress and the API is not considered to be frozen yet). [AL]
451 * Display more information on backtraces: source line & function names are
452 now displayed just like valgrind does (rely on addr2line tool) [MQ]
453 * New function: xbt_backtrace_display(). Sweet while debuging [MQ]
454 * Reworked a little bit some #include statements to load only required
455 headers. Some user code that relied on SimGrid to include stdlib or
456 stdio may need to include it by themselves. [AL]
457 * Fixed xbt/log.h. A missing SG_BEGIN_DECL prevented compilation with
459 * Renamed xbt_host_t into xbt_peer_t since it betterly describes what I
460 meant. This breaks the API of AMOK and of xbt/config. Sorry about this,
461 but I guess that almost nobody used those parts. [MQ]
463 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 14 Jul 2006 01:32:27 +0200
465 SimGrid (3.0.1) stable; urgency=low
468 * Unfortunately, I had missed 5 misnamed functions:
469 xbt_fifo_item_t xbt_fifo_newitem(void);
470 void xbt_fifo_freeitem(xbt_fifo_item_t);
471 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
472 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
473 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
474 They're now deprecated. Please use their new versions:
475 xbt_fifo_item_t xbt_fifo_new_item(void);
476 void xbt_fifo_free_item(xbt_fifo_item_t);
477 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
478 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
479 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
481 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
482 * Documentation: xbt_log module unmercifully reworked [MQ]
483 * Bugfix: there was a problem with the ending of contexts with
484 the pthread backend. It caused some weird deadlock or behavior
485 depending on the pthread implementation. [AL]
486 * Bugfix: get the exceptions raised in the simulator repport where
487 and why they come from when they are not catched in time [AL, MQ]
490 * Bugfix: Do repport the error when two non-connected hosts try to
491 exchange data (Thanks to Flavien for stumbling into this one) [AL]
494 * Add additionnal checkings on communications. Assert that two
495 communicating hosts are connected by a set of links... [AL]
498 * Add additionnal checkings on channel values in communication [AL]
499 * New: MSG_task_get_source to see on which host a task was generated [HC]
500 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
501 the number of tasks waiting to be received on channel and sent
503 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
504 waits for the first task coming from a given host.. [AL]
506 GRAS new functionnalities: [MQ]
507 * Enhance the parsing macro to allow the size of multidimentional objects
508 to be given thru annotations.
509 * New example (and documentation): Matrix Multiplication a la RPC
510 (as when I was young!) and fix a bunch of bugs found on the way.
512 GRAS performance improvements: [MQ]
514 * Reduce the amount of cbps creation/destruction by making it static to
515 datadesc_send/recv() and using a (newly created) cbps_reset (based on
518 * Change libdata to a set so that we can search for stuff by ID (and thus
519 reduce the insane amount of dict lookups)
522 * Actually implement gras_datadesc_copy() so that we don't have to mimick
523 RL communication on top of SG since it's so uneffective.
524 It may also be used for inter-thread communication in RL, one day.
525 * Use gras_datadesc_copy() to exchange messages on top of SG
527 - improve message exchange performance on top of SG
528 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
529 * Don't exchange on the network the size of the used part of buffer,
530 instead, specify the possible buffer size to read().
532 - reduces the amount of read/write calls (one pair per exchange)
533 - reduces the amount of exchanged data (the size)
534 - allows to retrieve all arrived data on receiver side, if we don't need
535 it right now (subsequent read will peek the buffer)
536 - allows the receiver to proceed with the begining of the stream before
537 everything is arrived
538 - make it possible to build an iov transport (using readv/writev)
540 - take care of the data with non-stable storage (like stacked data),
542 * If possible, TCP send uses vector I/O (when writev() is here)
543 - Don't use it for receive since we send data sizes and data on the
544 same stream, so we wouldn't be able to chain large amount of chunks
545 before having to flush the stuff to read the size.
546 * Rework the transport plugin mecanism to simplify it and reduce the
547 amount of pointer dereferencement when searching for the right function
550 * I guess that now, we do almost as few system calls as possible while
551 doing as few data copy as possible.
553 To improve it further, we could try to send all the sizes first and then
554 all the data (to use iov on receiving size), but it's only a partial
555 solution: when you have 2 dimensional data, the sizes of the second
556 dimension is data of the first dimension, so you need 3 streams.
558 I'm not sure the potential performance gains justify the coding burden.
560 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
562 SimGrid (3.00) stable; urgency=high
565 * New! Give the possibility to hijack the surf parser and thus bypass
566 MSG_create_environment and MSG_launch_application. Have a look at
567 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
569 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
571 SimGrid (2.96) unstable; urgency=low
576 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
577 This deprecates the xbt_error_t mecanisms.
578 It modifies (simplifies) all XBT and GRAS API.
579 MSG API keeps unchanged (exceptions raised by XBT are catched from
580 within MSG and masked with existing error handling facilities)
583 * New! Add a FATPIPE model. [AL]
584 * New! Add a parallel task model. [AL]
585 * New! Add automatically a loopback interface (in the default
586 network model) if none was precised.
589 * Bugfix: MSG_process_resume now works with the current running process.
591 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
592 * Modification of MSG_task_get_compute_duration. Once a task has been
593 processed, the value returned by this function is now equal to 0. [AL]
594 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
595 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
596 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
597 of a m_task. For now, MSG_task_cancel only works with computation
599 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
600 of the processor (in Mflop/s) regardless of the current load on the
602 * API Change: use proper naming convention for MSG_getClock and
603 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
605 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
606 This function changes the priority of a computation task. This priority
607 doesn't affect the transfer rate. A priority of 2 will make a task
608 receive two times more cpu power than the other ones. This function
609 has been added to suit the needs of Nguyen The Loc and hasn't been that
610 much tested yet. So if it fails, please report it and send me your code.
612 * API Change: removed all functions and types that were marked "deprecated"
613 since many months. Renamed MSG_global_init_args to MSG_global_init.
615 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
617 SimGrid (2.95) unstable; urgency=low
620 * Steal some nice code to GNU pth to fix context detection and usage [AL]
621 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
622 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
625 * Bugfix: Allow absolute paths for platform description files [MQ]
626 * Bugfix: do free the variables after use. Leads to drastic performance
628 * Implement max_duration (ie, timeouts) on resources [AL]
631 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
633 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
634 the needed support (timeouts) [AL]
637 * Implement measurement sockets. You can now get the bandwidth between two
638 hosts thanks to AMOK (see below). [MQ]
639 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
640 dynar over the network (yeah) [MQ]
641 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
642 * simple/static token-ring example. [Alexandre Colucci and MQ]
643 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
647 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
648 type parsing. "short" and "long" modifiers were also ignored; other
649 modifier (such as reference level) are still ignored. [MQ]
650 * Embeed the buffer size within the buffer itself on SG. [MQ]
651 That way, send() are atomic and cannot get intermixed anymore (at least
652 the ones which are less than 100k; bigger messages still have the issue)
653 * Array size pushed by the field, not by the field type (or each
654 and every long int will push stuff to the cbps) [MQ]
655 * use select() to sleep since it allows to portably sleep less than one
658 GRAS (minor cleanups)
659 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
662 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
663 * type category 'ignored' killed as it was never used and were difficult
665 * whether a type can cycle or not is now a flag, leaving room for more
666 flags (as "ignored", if we feel the need one day ;)
667 * Rename raw sockets to measurement sockets since "raw" has another
668 meaning in networking community.
671 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
672 and offers features not belonging to GRAS but that most applications
673 need. One day, it may be a set of plugins loadable at runtime.
674 * New module: bandwidth
675 bandwidth measurement between arbitrary nodes running this module. [MQ]
677 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
679 SimGrid (2.94) unstable; urgency=low
681 The first beta release of SimGrid 3 !
685 * Update the main page and the FAQ. Adding references to gforge.
688 * Add a gras_os_getpid function.
691 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
692 * Extend the logs so that they also print PID, hostname, date, ... if
694 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
695 and kill the old version which were in testsuite/
696 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
700 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
701 for users. Instead, each programming environment (GRAS, MSG,...) use it
702 when needed to provide such a feature to users.
703 Don't shortcut the mecanism or you will also shortcut the virtualization
704 you need on the simulator.
708 * Cleanups in configury with regard to compile optimization/warning flags.
709 Also add -fno-loop-optimize to any powerpc since it's the optimization
710 killing gcc (< 3.4.0).
711 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
712 needed by MSG examples complications
713 * Borrow configury beautifications from PHP
716 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
717 warning (thanks loris for stumbling into this one).
718 * Bugfix: stop loading private headers (gras_config.h) from the public
722 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
723 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
724 RL would imply a DNS resolver, which is *hard* to do in a portable way
725 (and therefore delayed).
726 * Implement a real timer mecanism and use it in timing macros. This allows
727 to avoid rounding errors and get a 0.000005 sec precision in timing
728 macros. While I was at it, various cleanups:
729 - allow to declare more than one timed section per file (fix a stupid bug)
730 - move some private declaration to the right place
731 - merge conditional execution and timing macros into emulation module
732 - document the module
733 - make sure the module cleanups its mess on gras_exit
734 * Documentation improvements:
735 - (new) how to compile applications using GRAS
736 - (new) emulation support (timing macros)
738 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
740 SimGrid (2.93) unstable; urgency=low
742 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
745 - Use Paje properly where used. Still to be sanitized properly.
746 - Portability fix: Add an implementation of the contexts using pthread
750 - Add xbt_procname(): returns the name of the current process.
751 Use it to show the current process's name in all logging.
753 - fix detection of older flex version and the reaction, since we do
754 depend on modern ones (we use lex_destroy)
755 - Better separation of SG and RL in the libs: remove all simulation code
756 from libgras. As a result, this lib is now only 200k when stripped.
757 Some of the xbt modules may also be duplicated (two sets and such) and
758 should be cleaned/killed before SG3.
759 - Insist on using xlC on AIX because of weird problems involving gcc there.
760 - Cleanup the make remote stuff. This is now done by scripts
761 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
762 mainly for our private use, but we're working on changing them to user
765 - Bugfix: flush the socket on close only if there is some *output*.
766 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
768 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
771 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
773 SimGrid (2.92) unstable; urgency=low
775 Alpha 3 on the path to SimGrid 3
779 - New! First try of benchmarking macros.
780 - New! First try so that gras_stub_generator generate deployment and
781 remote compilation helpers.
783 - Bugfix: Initialization fix in msg_test.
787 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
789 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
791 - API Change: gras_os_sleep to take the amount of seconds as a double.
792 Accepting an int was error prone since it was the only location where
793 seconds were coded as such. It leaded to damn rounding errors.
794 - Bugfix: Hard to belive that timers ever worked before this.
796 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
798 SimGrid (2.91) unstable; urgency=low
800 Alpha 2 on the path to SimGrid 3
804 - Bug fix in the lmm_solver.
806 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
807 through the function MSG_paje_output.
808 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
809 - It is possible to bound the rate of a communication in MSG with
810 MSG_task_put_bounded() (was already in the previous version but I had forgotten
811 to write it in the changelog).
812 - Bug fix to let GRAS run on top of MSG until we move it directly on top
817 - Various cleanups to the autotools stuff
818 - Begin to move Gras examples to examples/gras/
819 - Let make distcheck work again (yeah!)
821 - documentation overhauled using doxygen.
822 gtk-doc-tools is dead in SimGrid now.
823 - Automatically extract all existing logging categories, and add the list
824 to the documentation (long standing one, to say the less)
826 - Cleanup the known architecture table. Reorder the entries to group what
827 should be, and use a more consistent naming scheme.
828 (some of the test dataset are still to be regenerated)
829 - New! Allow library to register globals on each process just as userdata
831 This is implemented using a xbt_dict and not a xbt_set, so we loose the
832 lookup time (for now).
833 Use it in msg and trp.
834 This cleans a lot the internals and helps enforcing privacy of the
835 headers between the gras components.
836 - New! Add a timer mechanism, not unlike cron(8) and at(1).
837 - Bugfix: gras_os_time was delirious in RL.
838 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
840 - Reenable GRAS now that it works.
842 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
844 SimGrid (2.90) unstable; urgency=low
846 Alpha 1 on the path to SimGrid 3
848 * It is a long time since the last release of SimGrid. I'm sorry about
849 that but as I had told you, I was rewriting a lot of things. I apologize
850 to those who had been reporting bugs to me and that I had not answered.
851 If your bug is still in the new version, please tell me. Here is a
852 summary of the main changes.
854 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
855 lead by Martin Quinson. As a consequence SimGrid gains a lot in
856 portability, speed, and a lot more but you'll figure it out later.
857 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
858 I wanted to release the new MSG as soon as possible and I have
859 broken GRAS, which is the reason why, for now, only MSG is fully
860 functional. A laconic description of these projects is available
861 in the documentation.
863 * REVOLUTION 2: I have removed SG and I am now using a new simulation
864 kernel optimized for our needs (called SURF but only the developers
865 should use it). Hence, MSG is now roughly 30 times faster and I think
866 that by rewriting a little bit MSG, I could event speed it up a little
867 bit more. Beside the gain in speed, it is also much easier to encode a
868 new platform model with SURF than it was with SG. More to come...
870 * REVOLUTION 3: I have tried to change a little as possible the API of
871 MSG but a few things really had to disappear. The main differences
872 with the previous version are :
873 1) no more m_links_t and the corresponding functions. Platforms are
874 directly read from a XML description and cannot be hard-coded
875 anymore. The same format is used for application deployment
876 description. The new format is described in the documentation.
877 Have a look in tools/platform_generation. There is a tiny script
878 that converts from the old platform format to the new one. Concerning
879 the application deployment format, parsing the old one is tricky.
880 I think most of you should however be able to convert your files. If
881 it is really an issue, I can write a C code that does the conversion.
883 2) the toolbox tbx does not exist anymore. We now have a library
884 with much more data-structures but without the hash-tables (we have
885 dictionaries that are much faster).
887 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
889 *****************************************************************************
890 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
891 * don't feel like converting the oldies. (Mt) *
892 *****************************************************************************
895 Version 2.90: "the long awaited one"
896 - Finished rewriting and debugging MSG. Rewrote the documentation.
897 - disable GRAS for now since it needs to be ported to the newest SG
900 - Finish the port to windows (using mingw32 for cross-compile)
903 - Main loop and datastructures of SURF. A cpu resource object is
904 functional. Surf can thus be used to create cpu's with variable
905 performance on which you can execute some actions.
907 2004-11-15 Martin Quinson
908 - Port to ARM. Simply added the alignment and size descriptions. Should
909 work, but the ARM machines are so slow that I didn't had the opportunity
910 to 'make check' over there yet.
912 2004-11-15 Arnaud Legrand
913 - Trace manager now written. It uses a heap structure and is therefore
914 expected to be efficient. It may however be speeded up (particularly
915 when many events occur at the same date) by using red and black
916 trees. One day maybe...
917 - Max-min linear system solver written. It uses a sparse matrix
918 structure taking advantage of its expected use. Most operations are
919 O(1) and free/calloc are called as few as possible. The computation of
920 the minimum could however be improved by using a red and black tree
923 2004-11-03 Arnaud Legrand
924 - Rename every gras_* function that was in xbt/ to its xbt_
926 - Add a heap and a doubly-linked list to xbt
927 - Added a dichotomy to the dictionaries. make check works as well before
928 so I assume that the patch is correct. I do not know however if things
929 run effectively faster than before now. :)
931 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
932 SimGrid, and the version number is bumped to 2.x
934 2004-10-29 Martin Quinson
935 - Introduction of the remote errors.
936 They are the result of a RMI/RPC on the remote machine.
937 ErrCodes being scalar values, you can't get the host on which those
938 errors did happen. Extending the error mechanism as in Gnome is possible.
939 No idea yet whether it is a good idea.
941 2004-10-28 Martin Quinson
942 - Interface revolution: the Starred Structure Eradication.
943 I used to do typedef struct {} toto_t; and then handle *toto_t.
944 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
945 - ???_t is a valid type (builded with typedef)
946 - s_toto_t is a structure (access to fields with .)
947 - s_toto is a structure needing 'struct' keyword to be used
948 - e_toto_t is an enum
949 - toto_t is an 'object' (struct*)
951 typedef struct s_toto {} s_toto_t, *toto_t;
952 typedef enum {} e_toto_t;
953 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
954 s_toto_t) is private.
956 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
957 so that it can set the variable to NULL. It was so for dicts and sets,
958 it changed for dynars.
960 - Fix a bunch of memleaks in dict_remove
961 - Fix a bug in sg/server_socket opening: it failed all the time.
963 2004-10-07 Martin Quinson
964 - Speed up dynar lookup operation a bit.
966 gras_dynar_get is dead.
968 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
969 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
970 (which gives you the address of the stored data).
972 gras_dynar_get_as is an helpful macro which allows you to retrieve a
973 copy of the data using an affectation to do the job and not a memcpy.
975 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
976 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
978 It does not really speedup the dynar test because they are
979 setting elements all the time (and look them seldom). But the dict does
980 far more lookup than setting.
982 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
984 2004-10-05 Martin Quinson
985 - Allow to (en/dis)able the cycle detection at run time.
987 Whether we should check for cycle or not is now a property of each
988 datatype. When you think there may be some cycle, use datadesc_cycle_set.
989 datadesc_cycle_unset allow to remove this property when previously set.
991 Note that the cycle detection is off by default since it impacts the
992 performance. Watch the data you feed GRAS with ;)
994 This property is hereditary. Any element embedded in a structure having it
995 set have it set for the time of this data exchange.
997 You should set it both on sender and receiver side. If you don't set it on
998 sender side, it will enter an endless loop. If you forget on receiver
999 side, the cycles won't be recreated after communication.
1001 - Header reorganization.
1002 Kill gras_private.h, each submodule must load the headers it needs.
1004 2004-10-04 Martin Quinson
1005 - Interface revolution: do not try to survive to malloc failure.
1007 Now, gras_malloc and friends call gras_abort() on failure.
1008 As a conclusion, malloc_error is not a valid error anymore, and all
1009 functions for which it was the only gras_error_t return value are
1010 changed. They now return void, or there result directly.
1011 This simplify the API a lot.
1013 2004-09-29 Martin Quinson
1014 - Re-enable raw sockets.
1015 Created by gras_socket_{client,server}_ext;
1016 Used with gras_raw_{send,recv}
1019 It should allow to kill the last bits of gras first version soon.
1021 This is not completely satisfactory yet (duplicate code with
1022 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
1025 - Simplify transport plugin (internal) interface by not passing any
1026 argument to _server and _client, but embedding them in the socket
1029 2004-09-28 Martin Quinson
1030 - Finish the port to AIX.
1031 autoconf was my problem (segfault within the malloc replacement
1032 function. No idea why)
1034 2004-09-16 Martin Quinson
1035 - Fix some size_t madness on 64bit architectures.
1037 2004-09-08 Martin Quinson
1038 - Reduce the number of system headers loaded, overload some more system
1039 calls (such as malloc to cast the result of the system one, and work
1041 - Fix and reintroduce the config support
1043 2004-09-07 Martin Quinson
1044 - Source code reorganization to allow Arnaud to surf all over there.
1045 - Allow to document the logging categories.
1046 - Remove all uppercase from logging categories and useless cleanup in names.
1048 2004-08-18 Martin Quinson
1049 Version 0.6.2 (protocol not changed; API changed)
1050 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
1051 gras_error_t), and NULL when not found. Functions expecting a msgtype
1052 as argument (msg_wait; msg_send) deal with NULL argument by providing a
1053 hopefully usefull message.
1054 - Portability to prehistoric sparcs again
1056 2004-08-17 Martin Quinson
1057 Version 0.6.1 (protocol not changed; ABI not changed)
1058 - prealloc some buffers to speed things up
1060 2004-08-11 Martin Quinson
1061 Version 0.6 (protocol not changed; ABI expended)
1062 - The parsing macro can deal with the references, provided that you add
1063 the relevant annotations (using GRAS_ANNOTE(size,field_name))
1065 2004-08-09 Martin Quinson
1066 Version 0.5 (protocol not changed; ABI changed)
1067 - Allow to off turn the cycle detection code in data exchange at
1068 compilation time. It should be at run time, but I'm short of time (and
1069 the config stuff is still broken). That way, we keep dict out of the
1070 critical path, which is good because the performance is poor:
1071 - search not dichotomial yet
1072 - dynar give no way to access their content and memcpy everytime
1073 - In composed data description (struct, ref and so on), stop foolness of
1074 keeping the subtype's ID, but store the type itself. This keeps sets out
1075 of the critical path, which is good since they rely on dynar and
1076 dictionnaries. The only loose of that is that we cannot detect the
1077 redeclaration of a structure/union with another content (but I'm not sure
1078 the code detected well this error before anyway). We still can detect
1079 the redefinition discrepancy for the other types.
1080 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
1081 it breaks the code because of type-punning used all over the place).
1082 This breaks on all non-gcc architectures (for now).
1084 All those changes (plus the buffer of last time) allow me to gain 2 order
1085 of magnitude on cruel tests consisting of 800000 array of integers on two
1086 level of a hierarchical structure (200 secondes -> 4 secondes)
1089 - the selector of reference must now return the type it points to, not
1090 the ID of this type.
1092 2004-08-06 Martin Quinson
1093 Version 0.4 (protocol changed; ABI not changed)
1094 - Allow to pass --gras-log argument to processes in simulation mode. Really.
1095 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
1096 - Implement a buffer transport, and use it by default (it relies on tcp in
1097 real life and on sg in simulation).
1098 That's a bit hackish since I had a new field to the structure to store
1099 its data without interfering with the subtype ones. Inheritance
1100 is tricky in C. And that's a kind of reverse inheritance with one class
1101 derivating two classes. Or maybe a game with java interfaces. Anyway,
1102 that's damn hard in C (at least).
1103 Moreover, I got tired while trying to ensure plugin separation and
1104 genericity in SG mode. MSG wants me to do weird things, so let's go for
1105 cruel hacks (temporarily of course ;).
1106 See comment in transport_private.h:71
1107 - do not include all the _interface headers in private but in the files
1108 which really need them (to cut the compilation time when they are
1111 2004-07-26 Martin Quinson
1112 Version 0.3 (protocol not changed; ABI changed)
1113 - Major overhault of the datadesc interface to simplify it:
1114 - shorted the function names:
1115 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
1116 - add a trivial way to push/pop integers into the cbps without malloc.
1117 This allows to make really generic sub_type description, which simply
1118 pop their size of the stack.
1119 - add a function gras_datadesc_ref_pop_arr() which does what users want
1120 most of the time: Declare a dynamic array (which pops its size of the
1121 stack) and declare a reference to it. Poor name, but anyway.
1122 - kill the post-send callback, add a post-receive one
1124 2004-07-23 Martin Quinson
1125 Version 0.2 (protocol changed; ABI changed)
1126 - add some testing for cpbs in the test cases, and fix some more bugs.
1127 This invalidate again the little64 data file, since I cannot regenerate
1129 - remove an awfull optimization in the logging stuff, allowing me to:
1130 - understand it again
1131 - learn gcc how to check that the argument match the provided format
1132 - fix all errors revealed by gcc after that
1133 - internal keys of dict are not \0 terminated. Deal with it properly in
1134 loggings instead of segfaulting when the user want to see the logs :-/
1136 2004-07-22 Martin Quinson
1137 - Fix some stupid bug preventing cbps (callback postit) from working
1139 2004-07-21 Martin Quinson
1140 - Some documentation cleanups
1141 - remove the useless last argument of msgtype_declare
1142 - rename the Virtu functions to fit into the 'os' namespace
1143 - move headers src/include -> src/include/gras/ and stop fooling with
1144 gras -> . symbolic link
1145 - make distcheck is now successful
1147 2004-07-19 Martin Quinson
1149 - Build shared library also
1150 - Install html doc to the right location
1151 - stop removing maintainer files in make clean
1152 - build tests only on make check
1154 2004-07-13 Martin Quinson
1156 - No major issue in previous version => change versionning schema
1157 - Re-enable little64 convertion test now that Abdou kindly regenerated the
1158 corresponding dataset.
1160 2004-07-11 Martin Quinson
1162 - Get it working with any kind of structure (we can compute the padding
1163 bytes remotely for all the architectures I have access to)
1164 - Implement the structure parsing macro (still not quite robust/complete)
1165 - Improvement to the remote testing toysuite
1167 2004-07-10 Martin Quinson
1168 [autoconf mechanism]
1169 - get ride of a bunch of deprecated macros
1170 - actually run the test for two-compliment, not only compile it :-/
1171 - test whether the structures get packed (and bail out if yes. Damn.
1172 Alignment is a serious matter)
1173 - test whether the structures get compacted (but respecting the alignment
1174 constraints of each types)
1175 - test whether the array fields of structures can straddle alignment boundaries
1177 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
1178 (took me 2 hours to find that bug, looking at the wrong place)
1179 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
1180 before sending/receiving and are used to compute the offsets of fields
1181 - Given that padding size depend even on compiler options, keep track of
1182 alignment and aligned_size only for the current architecture. Not a big
1183 deal since we send structure fields one after the other (seems
1185 - Add the datastructure used for IEEE paper by the PBIO guys to the test
1186 program, let it work on linux/gcc/little32. portability todo.
1188 2004-07-08 Martin Quinson
1189 - import and improve remote compilation support from FAST
1190 - make sure make check works on half a dozen of machines out there
1192 2004-07-07 Martin Quinson
1193 Let's say it's version 0.0.3 ;)
1194 - Implement conversions (yuhu!)
1195 - Let it work on solaris (beside conversion, of course)
1196 - Stupid me, using rand() to generate the conversion datatests in not wise.
1198 2004-07-06 Martin Quinson
1199 - Let make dist work, since I'm gonna need it to compile on remote hosts
1200 - Let Tests/datadesc_usage write the architecture on which the file was
1201 generated as first byte.
1202 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
1203 architecture descriptions.
1204 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
1205 on those architectures.
1206 - Optimization: send/recv array of scalar in one shoot
1208 2004-07-05 Martin Quinson
1209 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
1211 - Plug a whole bunch of memleaks
1212 - each process now have to call gras_{init,exit}. One day, their log
1213 settings will be separated
1214 - Continue the code factorisation between SG, RL and common in Transport.
1216 2004-07-04 Martin Quinson
1218 - Redistribution between SG and RL.
1219 We wanna have to accept in SG, so move accepted related parts of RL in
1220 the common part. (more precisely, the dynar of all known sockets is no
1221 more a static in transport.c, but part of the process_data)
1223 [gras_stub_generator]
1224 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
1226 2004-07-03 Martin Quinson
1227 - Create a new log channel tbx containing dict, set, log, dynar (to shut
1228 them all up in one shot)
1230 - Fix the ugly case of reference to dynamic array.
1231 - New (semi-public) function gras_datadesc_size to allow the messaging
1232 layer to malloc the needed space for the buffer.
1234 - gras_socket_close now expect the socket to close (and not its address to
1235 put NULL in it after it). This is because the socket passed to handlers
1236 is one of their argument (=> not writable).
1238 - propagate the interface cleanup from last week in datadesc, ie remove a
1239 superfluous level of indirection. User pass adress of variable
1240 containing data (both when sending and receiving), and not of a variable
1241 being a pointer to the data. Let's say that I like it better ;)
1242 The price for that is constructs like "int msg=*(int*)payload" in
1243 handlers, but it's a fine price, IMHO.
1245 - Let it work in RL (yuhu)
1247 2004-06-21 Martin Quinson
1249 - porting SG plugin and SG select to new standards (works almost).
1250 - plug memleaks and fix bugs around.
1253 - cleanup the prototype of data recv and force users to specify when they
1254 want to handle references to objects. Test case working even for cycles.
1255 - plug memleaks. Valgrind is perfectly ok with this.
1257 2004-06-12 Martin Quinson
1259 - cleanup the separation between plugin and main code in plugin creation
1261 2004-06-11 Martin Quinson
1263 - Reput hook for raw sockets, needed for BW experiments
1264 - kill a few lines of dead code
1265 [Data description] Interface cleanup
1266 - gras_datadesc_by_name returns the searched type or NULL.
1267 That way, no variable is needed to use a type desc once, which makes
1269 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
1270 two parameters were strings (field name, type name), leading to
1272 [Dicos] Interface cleanup
1273 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
1274 This is consistant with the dynar API.
1276 2004-04-21 Martin Quinson
1278 - Porting to new standards.
1280 - interface cleanup.
1281 There is no bag anymore, no need to take extra provision to mask the
1282 pointers behind "ID".
1283 Better splitup of functions between files create/exchange/convert.
1284 This is still a bit artificial since convert and receive are so
1285 interleaved, but anyway.
1287 - add a queued message list to procdata (the ones not matching criteria
1289 - factorize some more code between SG and RL wrt procdata
1291 - use gras_exit in example to track memleaks
1292 - get rid of gs_example now that GS is properly integrated into gras
1293 - update run_test to integrate the lastest tests (datadesc)
1295 - rename WARNINGn macros to WARNn since it prooved error-prone
1297 2004-04-19 Martin Quinson
1299 - register init/exit functions within gras module mechanism
1300 - send/receive function.
1301 Convertion is not implemented, but short-cutted if not needed.
1302 struct/array elements are sent one by one (instead of block-wise), but
1303 nobody really cares (yet). Get a prototype before optimizing.
1304 - tests (using a file socket) for DD send/receive on:
1305 - base types: int, float
1306 - array: fixed size, string (ie ref to dynamic string)
1307 - structure: homogeneous, heterogeneous
1308 - chained list, graph with cycle
1309 Believe it or not, valgrind is not too unhappy with the results. The
1310 cycle happily segfaults, but the others are ok. And I'm sick of pointers
1314 - Bugfix when using a filename explicitely (instead of '-')
1316 2004-04-09 Martin Quinson
1318 - factorize more code between RL and SG in socket creation
1319 - Complete the implementation and tests of:
1321 o file (only in RL, and mainly for debugging)
1323 I lost 3 days to design a portable address resolver, and then decided
1324 that the prototype mainly have to run on my box.
1325 Addressing portability too early may be like optimizing too early :-/
1327 - use gras_init in the Tests instead of the crappy parse_log_opt
1328 (the latter function is removed)
1329 [Conditional execution]
1330 - New functions: gras_if_RL/gras_if_SG (basic support for this)
1331 [Code reorganisation]
1332 - Get rid of libgrasutils.a since it makes more trouble than it solves.
1333 Build examples against the RL library, since there is no way to disable
1334 its creation for now.
1336 For information, the beginning of coding on GRAS was back in june
1337 2003. I guess that every line has been rewritten at least twice since