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.
46 * Store smx_hosts in a dict since we only retrieve them by name
49 * Also include strbuff from xbt.h public header
50 * xbt_ex_display(): do not free the exception after displaying
51 This allows to do more with the given exception afterward.
52 Users should call xbt_ex_free() themselves.
55 * Massive internal cleanups:
56 - Store internal structures on processes instead of hosts (allows
57 to have more than one process per host, in addition of being more
59 - Cleanup the initialization/finalization process
60 - Kill a whole bunch of unneeded synchronization:
61 processes run in exclusive manner within the simulator
62 - Move queues from global tables to process data fields
64 - now accept -platform and -hostfile arguments
65 - Pass the right rank value to processes according on hostfile
68 * Do not require doxygen in maintainer mode
70 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr>
72 SimGrid (3.3) stable; urgency=high
76 * JAVA BINDINGS for MSG (you dreamt of them? We made them)
79 * Introduce the SIMIX module: factorize code between MSG and GRAS.
82 Until now, GRAS were using MSG as an interface to SURF. It was
83 quite difficult because both interface have several differences
84 (MSG channels vs GRAS sockets were the most notable point).
86 This also opens the gate to SMPI (which should occur soon) and speed
87 up simulations by to 40% (even if it were not the main goal).
89 **************************************
90 *DO NOT MIX 3.2 RESULTS WITH 3.3 ONES* Simix may changes simulations!
91 **************************************
92 The point is that events occuring at the exact same timestamp are
93 not scheduled in the same order with the old and new version. This
94 may be enough to completely change the execution of simulations in
95 some cases. Sorry for the inconvenience.
97 * Cleanup and upgrade the XML format to push further scalability
98 issues (check http://hal.inria.fr/inria-00256883/ for more info)
100 * Improve the testing infrastructure with tesh. Now a very large part of
101 the code is tested not only by being run but also by checking that the
102 output match an expected output [Mt].
104 * Move on to FleXML v1.7 for the embeeded XML parsers. This version
105 is really less memory-demanding, which should allow you to use
106 larger files in SimGrid [AL].
108 * Inform valgrind about our contextes, so that it becomes usable
109 with the default (and more effecient) version of SimGrid
110 [contributed by Sékou Diakite, many thanks]
113 * Introduce a listener thread in charge of receiving incomming
114 messages from the network. It allows to overlap communication and
115 computation but most notably, it removes some stupid deadlocks due
116 to the fact that so far, a process could not send and receive at
117 the same time. This made most non trivial communication schema
119 * Convert the PIDs from long int to int to match the MSG ones (and
121 * New function: gras_agent_spawn() to launch a new process on
122 current host. Only working in simulation for now. [Mt]
123 * New function: gras_os_hostport() returning a constant form (ie,
124 not needing to be freed) of "gras_os_hostname():gras_os_myport()"
127 * Make the backtrace of exceptions more human readable [Mt]
128 * New module: xbt/str [Mt]
129 a ton of string utility functions (split, join, printf to a newly
130 allocated buffer, trim, etc)
131 * New module: xbt/hash [Mt]
132 SHA1 hashing algorithm (more to come if needed)
133 * New module: xbt/synchro [Mt]
134 synchronization tools (mutex and conditions) working the same way
135 in simulation and in real life (mainly useful for GRAS, but not
137 * New module: xbt/queue [Mt]
138 classical producer/consumer synchronization scheme
139 * xbt_dynar_new_sync() creates a synchronized dynar. All access
140 (using the classical functions will get serialized) [Mt]
141 * Make dictionary internal table dynamic. No need to specify its size
142 anymore; functions xbt_dict_new_ext() and xbt_dict_hashsize_set()
144 * Make sure the log channels are organized as a tree under windows
145 (because of ANSI C compatibility issue, any channel were child of
149 * Cleaned many thing in surf and fixed a few bugs [AL].
150 * Add a nice command line configuration mechanism to compose models [AL].
151 * Add a new model for parallel tasks (ptask_L07) that is less buggy than
152 the previous one (KCCFLN05). It relies on something that looks like
153 a max-min sharing mechanism but cannot be written as such. A new solver
154 was thus designed [AL].
155 * Add a new solver to lmm. Based on Lagrange optimization and
156 gradient-based descent, it enables to efficiently maximise systems s.a
158 sum f_i(x_i) s.t Ax<= b with A_{i,j}>=0 and f_i a concave function.
160 This solver enables to propose two new network models for TCP Reno and
161 TCP Vegas based on Low's work. These models still need to be fully
162 tested though [Pedro Velho].
165 * Bug fix in SD_simulate. Now the time bound given as argument is
167 * Use the new parallel task model (ptask_L07) as default.
168 * Use the SURF command line configuration mechanism.
169 * 0-size tasks (for synchronization) should now work.
171 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Sun Apr 12 05:20:36 CEST 2009
173 SimGrid (3.2) stable; urgency=high
177 We still experience issues on this platform, but we believe that at
180 GRAS API BREAKAGE (for simplification purpose, sorry):
181 * the gras_msgtype_by_name is not used anymore. Instead of
182 gras_msg_send(toserver, gras_msgtype_by_name("request"), &request);
183 you can write (and must)
184 gras_msg_send(toserver, "request", &request);
185 - If you still want to pass a gras_msgtype_t to the function (to cache
186 the type and avoid the lookup time), use the gras_msg_send_() variant.
187 - Impacted functions:
188 gras_cb_register, gras_cb_unregister, gras_msg_send, gras_msg_wait,
189 gras_msg_rpccall, gras_msg_rpc_async_call, gras_msg_wait_ext
190 * The callbacks are now expected to return 0 when everything went well
191 (just like the main() function)
193 GRAS new features and improvements:
194 * New module mecanism where user code can use per process globals [Mt]
195 This is similar to gras_userdata_*() functions, but for libraries. It
196 factorize some code developped over and over in the examples and AMOK.
197 It has still to be documented and used (only amok/peermanagement is
199 * Fix a vicious bug in the TCP buffering mecanism which leaded to message
200 loss when they were small enough to fit into the buffer and sent quickly
201 enough so that they can all get received in one shoot.
202 * gras_datadesc_by_name and gras_msgtype_by_name: now raise an exception
203 if not found. Use the *_or_null() variant for the old semantic.
204 * In gras_msg_handle, do not discard messages without callback.
205 They are probably messages to be explicitly awaited later (ie, proofs of
206 mis-synchronization in userland since they are sent before being awaited)
208 * gras_socket_meas_send/recv: semantic changed!
209 The numerical arguments used to be (1) the total amount of data to send
210 and (2) msg_size. This was changed to (1) msg_size and (2) amount of
211 messages. This was need for the fool willing to send more than MAXINT
212 bytes on quite fat pipes.
215 * Do really rename the hostmanagement module to peermanagement. [Mt]
216 Ie, rename functions from amok_hm_* to amok_pm_*. This breaks the API,
217 but this is rather new and this was documented in the module
218 documentation (poor excuses, I admit)
219 * Bandwidth measurement semantic changed! This follows the changes to
220 gras_socket_meas_send/recv explained above.
223 * A sequential mode has been added to the workstations. When a workstation
224 is in sequential mode, it can execute only one task, and the other tasks
225 are waiting in a FIFO. [Christophe Thiery]
228 * The KCCFLN05 workstation model now handles parallel tasks. It is the
229 model for SIMDAG. [Christophe Thiery]
230 * Bug fix in the maxmin solver: Some values were close to 0 instead of
231 equal to 0, which caused some bad behaviors in
232 saturated_constraint_set_update. I now use a threshold mechanism like in
236 * When running manually src/testall, you select specific units [Mt]
237 testall is the result of our cunit mecanism, and should replace all
238 the scripty thingy around since bash don't run easily on billware.
240 * A mallocator system has been added. [Christophe Thiery]
241 Mallocators allow you to recycle your unused objects instead of freeing them
242 and allocating new ones.
244 Documentation update:
245 * FAQ reworking + New FAQs:
246 - "Valgrind spits tons of errors!" [Mt]
247 - "How to repport bugs" [Mt]
248 - "Cross-compiling a Windows DLL of SimGrid from Linux" [Mt]
249 - "What is the difference between MSG, SimDag, and GRAS?" [Mt]
250 - Communication time measurement within MSG [AL]
251 - I experience weird communication times when I change the latency [AL]
254 - an introduction to the framework and to the used communication model
255 - an initiatic tour introducing the most proheminent features:
257 . Lesson 0: Installing GRAS
258 . Lesson 1: Setting up your own project
259 o Part 2: Message passing
260 . Lesson 2: Exchanging simple messages
261 . Lesson 3: Passing arguments to the processes (in SG)
262 . Lesson 4: Attaching callbacks to messages
263 . Lesson 5: Using globals in processes
264 . Lesson 6: Logging informations properly
265 . Lesson 7: Using internal timers
266 . Lesson 8: Handling errors through exceptions
267 . Lesson 9: Exchanging simple data
268 . Lesson 10: Remote Procedure Calling (RPC)
269 . Lesson 11: Explicitely waiting for messages
270 . Recapping of message passing features in GRAS
271 - A HOWTO section containing:
272 o HOWTO design a GRAS application
273 More are due, of course. They will come latter. In the meanwhile, you can
274 check the examples which are still here.
276 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri Mar 16 21:11:46 CET 2007
278 SimGrid (3.1) stable; urgency=high
282 There was a stack corruption somewhere, visible only when optimizing
283 with these versions. [Vince]
286 * This is a NEW module! SimDAG (SD for short) is a revival of the old SG
287 module that enabled to play with Directed Acyclic Graphs. It is built
288 directly on top of SURF and provides an API rather close to the old
289 SG. Some old codes using SG are currently under rewrite to check that
290 all needful functions are provided. [Christophe Thiery]
293 * Complete rewrite of the KCCFLN05 workstation model. It is now an
294 extension of the classical CLM03 model that gracefully handles
295 failures. This is now the default model for MSG and GRAS. It doesn't
296 handle parallel tasks yet though. [AL]
297 * Bug fix: Weights were not correctly set in the network part.
298 WARNING: This may have resulted in incorrect results with simulations
299 where there are more than one flow on a given link. [AL]
302 * After a (long ?) discussion on simgrid-devel, we have decided that the
303 convention we had on units was stupid. That is why it has been decided
304 to move from (MBits, MFlops, seconds) to (Bits, Flops, seconds).
305 WARNING : This means that all previous platform files will not work as
306 such with this version! A warning is issued to ask users to update
308 A conversion script can be found in the contrib module of the CVS, under
309 the name contrib/platform_generation/surfxml_update.pl [MQ]
312 * Bug fix: Processes were started in reverse order, wrt deployment file.
313 WARNING: if your code relies on this bug, please fix it. [AL]
314 * Bug fix: Add a test in MSG_task_execute to stop whenever a task is
315 being executed on two different locations. [AL]
316 * Bug fix: Failures are now better supported thanks to Derrick's tests
317 (there was many failure situations I hadn't thought of and that weren't
318 correctly handled). [AL]
319 * New function: MSG_host_is_avail indicates you whether a given m_host_t
323 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
324 Exception occurring on server-side are propagated back to client (!).
326 API CHANGE: the callback changed their prototype. Change:
327 int my_handler(gras_socket_t expeditor, void *payload_data) {
329 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
330 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
332 * New! function: gras_msg_handleall to deal with all messages arriving
333 within a given period.
334 * New! function: gras_socket_server_range to get a server socket in a
335 range of port numbers (ease to avoid port number conflicts) [MQ]
336 * New! gras processes display their backtrace when they get a SIGUSR1
337 or when Ctrl-C is pressed. Use Ctrl-C Ctrl-C to exit.
338 Sweet to debug RL processes [MQ]
342 - Do not force experiment sizes to be expressed in kb, or it becomes
343 impossible to measure the latency this way (needs one byte-long tests)
344 WARNING: this changes the amok_bw_* function semantic. [MQ]
345 - Implements the link saturation stuff. [MQ]
346 * Peer management module:
347 New! module factorizing code that we wrote over and over [MQ].
350 * New module: cunit (my jUnit implementation in ansi C) [MQ]
351 - Test units are placed directly into the library code, they get extracted
352 automatically and placed into the src/testall binary.
353 - Convert most of the XBT tests to this system.
354 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
355 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
356 live in the same C code. This required to change a little bit the API
357 of surfxml parsing but shouldn't be an issue for end-users. [AL]
358 * New module: sparse graph structure with basic algorithms (this is work
359 in progress and the API is not considered to be frozen yet). [AL]
360 * Display more information on backtraces: source line & function names are
361 now displayed just like valgrind does (rely on addr2line tool) [MQ]
362 * New function: xbt_backtrace_display(). Sweet while debuging [MQ]
363 * Reworked a little bit some #include statements to load only required
364 headers. Some user code that relied on SimGrid to include stdlib or
365 stdio may need to include it by themselves. [AL]
366 * Fixed xbt/log.h. A missing SG_BEGIN_DECL prevented compilation with
368 * Renamed xbt_host_t into xbt_peer_t since it betterly describes what I
369 meant. This breaks the API of AMOK and of xbt/config. Sorry about this,
370 but I guess that almost nobody used those parts. [MQ]
372 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 14 Jul 2006 01:32:27 +0200
374 SimGrid (3.0.1) stable; urgency=low
377 * Unfortunately, I had missed 5 misnamed functions:
378 xbt_fifo_item_t xbt_fifo_newitem(void);
379 void xbt_fifo_freeitem(xbt_fifo_item_t);
380 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
381 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
382 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
383 They're now deprecated. Please use their new versions:
384 xbt_fifo_item_t xbt_fifo_new_item(void);
385 void xbt_fifo_free_item(xbt_fifo_item_t);
386 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
387 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
388 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
390 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
391 * Documentation: xbt_log module unmercifully reworked [MQ]
392 * Bugfix: there was a problem with the ending of contexts with
393 the pthread backend. It caused some weird deadlock or behavior
394 depending on the pthread implementation. [AL]
395 * Bugfix: get the exceptions raised in the simulator repport where
396 and why they come from when they are not catched in time [AL, MQ]
399 * Bugfix: Do repport the error when two non-connected hosts try to
400 exchange data (Thanks to Flavien for stumbling into this one) [AL]
403 * Add additionnal checkings on communications. Assert that two
404 communicating hosts are connected by a set of links... [AL]
407 * Add additionnal checkings on channel values in communication [AL]
408 * New: MSG_task_get_source to see on which host a task was generated [HC]
409 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
410 the number of tasks waiting to be received on channel and sent
412 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
413 waits for the first task coming from a given host.. [AL]
415 GRAS new functionnalities: [MQ]
416 * Enhance the parsing macro to allow the size of multidimentional objects
417 to be given thru annotations.
418 * New example (and documentation): Matrix Multiplication a la RPC
419 (as when I was young!) and fix a bunch of bugs found on the way.
421 GRAS performance improvements: [MQ]
423 * Reduce the amount of cbps creation/destruction by making it static to
424 datadesc_send/recv() and using a (newly created) cbps_reset (based on
427 * Change libdata to a set so that we can search for stuff by ID (and thus
428 reduce the insane amount of dict lookups)
431 * Actually implement gras_datadesc_copy() so that we don't have to mimick
432 RL communication on top of SG since it's so uneffective.
433 It may also be used for inter-thread communication in RL, one day.
434 * Use gras_datadesc_copy() to exchange messages on top of SG
436 - improve message exchange performance on top of SG
437 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
438 * Don't exchange on the network the size of the used part of buffer,
439 instead, specify the possible buffer size to read().
441 - reduces the amount of read/write calls (one pair per exchange)
442 - reduces the amount of exchanged data (the size)
443 - allows to retrieve all arrived data on receiver side, if we don't need
444 it right now (subsequent read will peek the buffer)
445 - allows the receiver to proceed with the begining of the stream before
446 everything is arrived
447 - make it possible to build an iov transport (using readv/writev)
449 - take care of the data with non-stable storage (like stacked data),
451 * If possible, TCP send uses vector I/O (when writev() is here)
452 - Don't use it for receive since we send data sizes and data on the
453 same stream, so we wouldn't be able to chain large amount of chunks
454 before having to flush the stuff to read the size.
455 * Rework the transport plugin mecanism to simplify it and reduce the
456 amount of pointer dereferencement when searching for the right function
459 * I guess that now, we do almost as few system calls as possible while
460 doing as few data copy as possible.
462 To improve it further, we could try to send all the sizes first and then
463 all the data (to use iov on receiving size), but it's only a partial
464 solution: when you have 2 dimensional data, the sizes of the second
465 dimension is data of the first dimension, so you need 3 streams.
467 I'm not sure the potential performance gains justify the coding burden.
469 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
471 SimGrid (3.00) stable; urgency=high
474 * New! Give the possibility to hijack the surf parser and thus bypass
475 MSG_create_environment and MSG_launch_application. Have a look at
476 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
478 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
480 SimGrid (2.96) unstable; urgency=low
485 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
486 This deprecates the xbt_error_t mecanisms.
487 It modifies (simplifies) all XBT and GRAS API.
488 MSG API keeps unchanged (exceptions raised by XBT are catched from
489 within MSG and masked with existing error handling facilities)
492 * New! Add a FATPIPE model. [AL]
493 * New! Add a parallel task model. [AL]
494 * New! Add automatically a loopback interface (in the default
495 network model) if none was precised.
498 * Bugfix: MSG_process_resume now works with the current running process.
500 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
501 * Modification of MSG_task_get_compute_duration. Once a task has been
502 processed, the value returned by this function is now equal to 0. [AL]
503 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
504 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
505 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
506 of a m_task. For now, MSG_task_cancel only works with computation
508 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
509 of the processor (in Mflop/s) regardless of the current load on the
511 * API Change: use proper naming convention for MSG_getClock and
512 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
514 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
515 This function changes the priority of a computation task. This priority
516 doesn't affect the transfer rate. A priority of 2 will make a task
517 receive two times more cpu power than the other ones. This function
518 has been added to suit the needs of Nguyen The Loc and hasn't been that
519 much tested yet. So if it fails, please report it and send me your code.
521 * API Change: removed all functions and types that were marked "deprecated"
522 since many months. Renamed MSG_global_init_args to MSG_global_init.
524 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
526 SimGrid (2.95) unstable; urgency=low
529 * Steal some nice code to GNU pth to fix context detection and usage [AL]
530 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
531 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
534 * Bugfix: Allow absolute paths for platform description files [MQ]
535 * Bugfix: do free the variables after use. Leads to drastic performance
537 * Implement max_duration (ie, timeouts) on resources [AL]
540 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
542 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
543 the needed support (timeouts) [AL]
546 * Implement measurement sockets. You can now get the bandwidth between two
547 hosts thanks to AMOK (see below). [MQ]
548 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
549 dynar over the network (yeah) [MQ]
550 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
551 * simple/static token-ring example. [Alexandre Colucci and MQ]
552 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
556 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
557 type parsing. "short" and "long" modifiers were also ignored; other
558 modifier (such as reference level) are still ignored. [MQ]
559 * Embeed the buffer size within the buffer itself on SG. [MQ]
560 That way, send() are atomic and cannot get intermixed anymore (at least
561 the ones which are less than 100k; bigger messages still have the issue)
562 * Array size pushed by the field, not by the field type (or each
563 and every long int will push stuff to the cbps) [MQ]
564 * use select() to sleep since it allows to portably sleep less than one
567 GRAS (minor cleanups)
568 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
571 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
572 * type category 'ignored' killed as it was never used and were difficult
574 * whether a type can cycle or not is now a flag, leaving room for more
575 flags (as "ignored", if we feel the need one day ;)
576 * Rename raw sockets to measurement sockets since "raw" has another
577 meaning in networking community.
580 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
581 and offers features not belonging to GRAS but that most applications
582 need. One day, it may be a set of plugins loadable at runtime.
583 * New module: bandwidth
584 bandwidth measurement between arbitrary nodes running this module. [MQ]
586 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
588 SimGrid (2.94) unstable; urgency=low
590 The first beta release of SimGrid 3 !
594 * Update the main page and the FAQ. Adding references to gforge.
597 * Add a gras_os_getpid function.
600 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
601 * Extend the logs so that they also print PID, hostname, date, ... if
603 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
604 and kill the old version which were in testsuite/
605 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
609 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
610 for users. Instead, each programming environment (GRAS, MSG,...) use it
611 when needed to provide such a feature to users.
612 Don't shortcut the mecanism or you will also shortcut the virtualization
613 you need on the simulator.
617 * Cleanups in configury with regard to compile optimization/warning flags.
618 Also add -fno-loop-optimize to any powerpc since it's the optimization
619 killing gcc (< 3.4.0).
620 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
621 needed by MSG examples complications
622 * Borrow configury beautifications from PHP
625 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
626 warning (thanks loris for stumbling into this one).
627 * Bugfix: stop loading private headers (gras_config.h) from the public
631 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
632 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
633 RL would imply a DNS resolver, which is *hard* to do in a portable way
634 (and therefore delayed).
635 * Implement a real timer mecanism and use it in timing macros. This allows
636 to avoid rounding errors and get a 0.000005 sec precision in timing
637 macros. While I was at it, various cleanups:
638 - allow to declare more than one timed section per file (fix a stupid bug)
639 - move some private declaration to the right place
640 - merge conditional execution and timing macros into emulation module
641 - document the module
642 - make sure the module cleanups its mess on gras_exit
643 * Documentation improvements:
644 - (new) how to compile applications using GRAS
645 - (new) emulation support (timing macros)
647 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
649 SimGrid (2.93) unstable; urgency=low
651 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
654 - Use Paje properly where used. Still to be sanitized properly.
655 - Portability fix: Add an implementation of the contexts using pthread
659 - Add xbt_procname(): returns the name of the current process.
660 Use it to show the current process's name in all logging.
662 - fix detection of older flex version and the reaction, since we do
663 depend on modern ones (we use lex_destroy)
664 - Better separation of SG and RL in the libs: remove all simulation code
665 from libgras. As a result, this lib is now only 200k when stripped.
666 Some of the xbt modules may also be duplicated (two sets and such) and
667 should be cleaned/killed before SG3.
668 - Insist on using xlC on AIX because of weird problems involving gcc there.
669 - Cleanup the make remote stuff. This is now done by scripts
670 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
671 mainly for our private use, but we're working on changing them to user
674 - Bugfix: flush the socket on close only if there is some *output*.
675 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
677 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
680 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
682 SimGrid (2.92) unstable; urgency=low
684 Alpha 3 on the path to SimGrid 3
688 - New! First try of benchmarking macros.
689 - New! First try so that gras_stub_generator generate deployment and
690 remote compilation helpers.
692 - Bugfix: Initialization fix in msg_test.
696 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
698 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
700 - API Change: gras_os_sleep to take the amount of seconds as a double.
701 Accepting an int was error prone since it was the only location where
702 seconds were coded as such. It leaded to damn rounding errors.
703 - Bugfix: Hard to belive that timers ever worked before this.
705 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
707 SimGrid (2.91) unstable; urgency=low
709 Alpha 2 on the path to SimGrid 3
713 - Bug fix in the lmm_solver.
715 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
716 through the function MSG_paje_output.
717 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
718 - It is possible to bound the rate of a communication in MSG with
719 MSG_task_put_bounded() (was already in the previous version but I had forgotten
720 to write it in the changelog).
721 - Bug fix to let GRAS run on top of MSG until we move it directly on top
726 - Various cleanups to the autotools stuff
727 - Begin to move Gras examples to examples/gras/
728 - Let make distcheck work again (yeah!)
730 - documentation overhauled using doxygen.
731 gtk-doc-tools is dead in SimGrid now.
732 - Automatically extract all existing logging categories, and add the list
733 to the documentation (long standing one, to say the less)
735 - Cleanup the known architecture table. Reorder the entries to group what
736 should be, and use a more consistent naming scheme.
737 (some of the test dataset are still to be regenerated)
738 - New! Allow library to register globals on each process just as userdata
740 This is implemented using a xbt_dict and not a xbt_set, so we loose the
741 lookup time (for now).
742 Use it in msg and trp.
743 This cleans a lot the internals and helps enforcing privacy of the
744 headers between the gras components.
745 - New! Add a timer mechanism, not unlike cron(8) and at(1).
746 - Bugfix: gras_os_time was delirious in RL.
747 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
749 - Reenable GRAS now that it works.
751 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
753 SimGrid (2.90) unstable; urgency=low
755 Alpha 1 on the path to SimGrid 3
757 * It is a long time since the last release of SimGrid. I'm sorry about
758 that but as I had told you, I was rewriting a lot of things. I apologize
759 to those who had been reporting bugs to me and that I had not answered.
760 If your bug is still in the new version, please tell me. Here is a
761 summary of the main changes.
763 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
764 lead by Martin Quinson. As a consequence SimGrid gains a lot in
765 portability, speed, and a lot more but you'll figure it out later.
766 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
767 I wanted to release the new MSG as soon as possible and I have
768 broken GRAS, which is the reason why, for now, only MSG is fully
769 functional. A laconic description of these projects is available
770 in the documentation.
772 * REVOLUTION 2: I have removed SG and I am now using a new simulation
773 kernel optimized for our needs (called SURF but only the developers
774 should use it). Hence, MSG is now roughly 30 times faster and I think
775 that by rewriting a little bit MSG, I could event speed it up a little
776 bit more. Beside the gain in speed, it is also much easier to encode a
777 new platform model with SURF than it was with SG. More to come...
779 * REVOLUTION 3: I have tried to change a little as possible the API of
780 MSG but a few things really had to disappear. The main differences
781 with the previous version are :
782 1) no more m_links_t and the corresponding functions. Platforms are
783 directly read from a XML description and cannot be hard-coded
784 anymore. The same format is used for application deployment
785 description. The new format is described in the documentation.
786 Have a look in tools/platform_generation. There is a tiny script
787 that converts from the old platform format to the new one. Concerning
788 the application deployment format, parsing the old one is tricky.
789 I think most of you should however be able to convert your files. If
790 it is really an issue, I can write a C code that does the conversion.
792 2) the toolbox tbx does not exist anymore. We now have a library
793 with much more data-structures but without the hash-tables (we have
794 dictionaries that are much faster).
796 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
798 *****************************************************************************
799 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
800 * don't feel like converting the oldies. (Mt) *
801 *****************************************************************************
804 Version 2.90: "the long awaited one"
805 - Finished rewriting and debugging MSG. Rewrote the documentation.
806 - disable GRAS for now since it needs to be ported to the newest SG
809 - Finish the port to windows (using mingw32 for cross-compile)
812 - Main loop and datastructures of SURF. A cpu resource object is
813 functional. Surf can thus be used to create cpu's with variable
814 performance on which you can execute some actions.
816 2004-11-15 Martin Quinson
817 - Port to ARM. Simply added the alignment and size descriptions. Should
818 work, but the ARM machines are so slow that I didn't had the opportunity
819 to 'make check' over there yet.
821 2004-11-15 Arnaud Legrand
822 - Trace manager now written. It uses a heap structure and is therefore
823 expected to be efficient. It may however be speeded up (particularly
824 when many events occur at the same date) by using red and black
825 trees. One day maybe...
826 - Max-min linear system solver written. It uses a sparse matrix
827 structure taking advantage of its expected use. Most operations are
828 O(1) and free/calloc are called as few as possible. The computation of
829 the minimum could however be improved by using a red and black tree
832 2004-11-03 Arnaud Legrand
833 - Rename every gras_* function that was in xbt/ to its xbt_
835 - Add a heap and a doubly-linked list to xbt
836 - Added a dichotomy to the dictionaries. make check works as well before
837 so I assume that the patch is correct. I do not know however if things
838 run effectively faster than before now. :)
840 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
841 SimGrid, and the version number is bumped to 2.x
843 2004-10-29 Martin Quinson
844 - Introduction of the remote errors.
845 They are the result of a RMI/RPC on the remote machine.
846 ErrCodes being scalar values, you can't get the host on which those
847 errors did happen. Extending the error mechanism as in Gnome is possible.
848 No idea yet whether it is a good idea.
850 2004-10-28 Martin Quinson
851 - Interface revolution: the Starred Structure Eradication.
852 I used to do typedef struct {} toto_t; and then handle *toto_t.
853 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
854 - ???_t is a valid type (builded with typedef)
855 - s_toto_t is a structure (access to fields with .)
856 - s_toto is a structure needing 'struct' keyword to be used
857 - e_toto_t is an enum
858 - toto_t is an 'object' (struct*)
860 typedef struct s_toto {} s_toto_t, *toto_t;
861 typedef enum {} e_toto_t;
862 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
863 s_toto_t) is private.
865 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
866 so that it can set the variable to NULL. It was so for dicts and sets,
867 it changed for dynars.
869 - Fix a bunch of memleaks in dict_remove
870 - Fix a bug in sg/server_socket opening: it failed all the time.
872 2004-10-07 Martin Quinson
873 - Speed up dynar lookup operation a bit.
875 gras_dynar_get is dead.
877 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
878 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
879 (which gives you the address of the stored data).
881 gras_dynar_get_as is an helpful macro which allows you to retrieve a
882 copy of the data using an affectation to do the job and not a memcpy.
884 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
885 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
887 It does not really speedup the dynar test because they are
888 setting elements all the time (and look them seldom). But the dict does
889 far more lookup than setting.
891 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
893 2004-10-05 Martin Quinson
894 - Allow to (en/dis)able the cycle detection at run time.
896 Whether we should check for cycle or not is now a property of each
897 datatype. When you think there may be some cycle, use datadesc_cycle_set.
898 datadesc_cycle_unset allow to remove this property when previously set.
900 Note that the cycle detection is off by default since it impacts the
901 performance. Watch the data you feed GRAS with ;)
903 This property is hereditary. Any element embedded in a structure having it
904 set have it set for the time of this data exchange.
906 You should set it both on sender and receiver side. If you don't set it on
907 sender side, it will enter an endless loop. If you forget on receiver
908 side, the cycles won't be recreated after communication.
910 - Header reorganization.
911 Kill gras_private.h, each submodule must load the headers it needs.
913 2004-10-04 Martin Quinson
914 - Interface revolution: do not try to survive to malloc failure.
916 Now, gras_malloc and friends call gras_abort() on failure.
917 As a conclusion, malloc_error is not a valid error anymore, and all
918 functions for which it was the only gras_error_t return value are
919 changed. They now return void, or there result directly.
920 This simplify the API a lot.
922 2004-09-29 Martin Quinson
923 - Re-enable raw sockets.
924 Created by gras_socket_{client,server}_ext;
925 Used with gras_raw_{send,recv}
928 It should allow to kill the last bits of gras first version soon.
930 This is not completely satisfactory yet (duplicate code with
931 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
934 - Simplify transport plugin (internal) interface by not passing any
935 argument to _server and _client, but embedding them in the socket
938 2004-09-28 Martin Quinson
939 - Finish the port to AIX.
940 autoconf was my problem (segfault within the malloc replacement
941 function. No idea why)
943 2004-09-16 Martin Quinson
944 - Fix some size_t madness on 64bit architectures.
946 2004-09-08 Martin Quinson
947 - Reduce the number of system headers loaded, overload some more system
948 calls (such as malloc to cast the result of the system one, and work
950 - Fix and reintroduce the config support
952 2004-09-07 Martin Quinson
953 - Source code reorganization to allow Arnaud to surf all over there.
954 - Allow to document the logging categories.
955 - Remove all uppercase from logging categories and useless cleanup in names.
957 2004-08-18 Martin Quinson
958 Version 0.6.2 (protocol not changed; API changed)
959 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
960 gras_error_t), and NULL when not found. Functions expecting a msgtype
961 as argument (msg_wait; msg_send) deal with NULL argument by providing a
962 hopefully usefull message.
963 - Portability to prehistoric sparcs again
965 2004-08-17 Martin Quinson
966 Version 0.6.1 (protocol not changed; ABI not changed)
967 - prealloc some buffers to speed things up
969 2004-08-11 Martin Quinson
970 Version 0.6 (protocol not changed; ABI expended)
971 - The parsing macro can deal with the references, provided that you add
972 the relevant annotations (using GRAS_ANNOTE(size,field_name))
974 2004-08-09 Martin Quinson
975 Version 0.5 (protocol not changed; ABI changed)
976 - Allow to off turn the cycle detection code in data exchange at
977 compilation time. It should be at run time, but I'm short of time (and
978 the config stuff is still broken). That way, we keep dict out of the
979 critical path, which is good because the performance is poor:
980 - search not dichotomial yet
981 - dynar give no way to access their content and memcpy everytime
982 - In composed data description (struct, ref and so on), stop foolness of
983 keeping the subtype's ID, but store the type itself. This keeps sets out
984 of the critical path, which is good since they rely on dynar and
985 dictionnaries. The only loose of that is that we cannot detect the
986 redeclaration of a structure/union with another content (but I'm not sure
987 the code detected well this error before anyway). We still can detect
988 the redefinition discrepancy for the other types.
989 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
990 it breaks the code because of type-punning used all over the place).
991 This breaks on all non-gcc architectures (for now).
993 All those changes (plus the buffer of last time) allow me to gain 2 order
994 of magnitude on cruel tests consisting of 800000 array of integers on two
995 level of a hierarchical structure (200 secondes -> 4 secondes)
998 - the selector of reference must now return the type it points to, not
1001 2004-08-06 Martin Quinson
1002 Version 0.4 (protocol changed; ABI not changed)
1003 - Allow to pass --gras-log argument to processes in simulation mode. Really.
1004 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
1005 - Implement a buffer transport, and use it by default (it relies on tcp in
1006 real life and on sg in simulation).
1007 That's a bit hackish since I had a new field to the structure to store
1008 its data without interfering with the subtype ones. Inheritance
1009 is tricky in C. And that's a kind of reverse inheritance with one class
1010 derivating two classes. Or maybe a game with java interfaces. Anyway,
1011 that's damn hard in C (at least).
1012 Moreover, I got tired while trying to ensure plugin separation and
1013 genericity in SG mode. MSG wants me to do weird things, so let's go for
1014 cruel hacks (temporarily of course ;).
1015 See comment in transport_private.h:71
1016 - do not include all the _interface headers in private but in the files
1017 which really need them (to cut the compilation time when they are
1020 2004-07-26 Martin Quinson
1021 Version 0.3 (protocol not changed; ABI changed)
1022 - Major overhault of the datadesc interface to simplify it:
1023 - shorted the function names:
1024 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
1025 - add a trivial way to push/pop integers into the cbps without malloc.
1026 This allows to make really generic sub_type description, which simply
1027 pop their size of the stack.
1028 - add a function gras_datadesc_ref_pop_arr() which does what users want
1029 most of the time: Declare a dynamic array (which pops its size of the
1030 stack) and declare a reference to it. Poor name, but anyway.
1031 - kill the post-send callback, add a post-receive one
1033 2004-07-23 Martin Quinson
1034 Version 0.2 (protocol changed; ABI changed)
1035 - add some testing for cpbs in the test cases, and fix some more bugs.
1036 This invalidate again the little64 data file, since I cannot regenerate
1038 - remove an awfull optimization in the logging stuff, allowing me to:
1039 - understand it again
1040 - learn gcc how to check that the argument match the provided format
1041 - fix all errors revealed by gcc after that
1042 - internal keys of dict are not \0 terminated. Deal with it properly in
1043 loggings instead of segfaulting when the user want to see the logs :-/
1045 2004-07-22 Martin Quinson
1046 - Fix some stupid bug preventing cbps (callback postit) from working
1048 2004-07-21 Martin Quinson
1049 - Some documentation cleanups
1050 - remove the useless last argument of msgtype_declare
1051 - rename the Virtu functions to fit into the 'os' namespace
1052 - move headers src/include -> src/include/gras/ and stop fooling with
1053 gras -> . symbolic link
1054 - make distcheck is now successful
1056 2004-07-19 Martin Quinson
1058 - Build shared library also
1059 - Install html doc to the right location
1060 - stop removing maintainer files in make clean
1061 - build tests only on make check
1063 2004-07-13 Martin Quinson
1065 - No major issue in previous version => change versionning schema
1066 - Re-enable little64 convertion test now that Abdou kindly regenerated the
1067 corresponding dataset.
1069 2004-07-11 Martin Quinson
1071 - Get it working with any kind of structure (we can compute the padding
1072 bytes remotely for all the architectures I have access to)
1073 - Implement the structure parsing macro (still not quite robust/complete)
1074 - Improvement to the remote testing toysuite
1076 2004-07-10 Martin Quinson
1077 [autoconf mechanism]
1078 - get ride of a bunch of deprecated macros
1079 - actually run the test for two-compliment, not only compile it :-/
1080 - test whether the structures get packed (and bail out if yes. Damn.
1081 Alignment is a serious matter)
1082 - test whether the structures get compacted (but respecting the alignment
1083 constraints of each types)
1084 - test whether the array fields of structures can straddle alignment boundaries
1086 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
1087 (took me 2 hours to find that bug, looking at the wrong place)
1088 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
1089 before sending/receiving and are used to compute the offsets of fields
1090 - Given that padding size depend even on compiler options, keep track of
1091 alignment and aligned_size only for the current architecture. Not a big
1092 deal since we send structure fields one after the other (seems
1094 - Add the datastructure used for IEEE paper by the PBIO guys to the test
1095 program, let it work on linux/gcc/little32. portability todo.
1097 2004-07-08 Martin Quinson
1098 - import and improve remote compilation support from FAST
1099 - make sure make check works on half a dozen of machines out there
1101 2004-07-07 Martin Quinson
1102 Let's say it's version 0.0.3 ;)
1103 - Implement conversions (yuhu!)
1104 - Let it work on solaris (beside conversion, of course)
1105 - Stupid me, using rand() to generate the conversion datatests in not wise.
1107 2004-07-06 Martin Quinson
1108 - Let make dist work, since I'm gonna need it to compile on remote hosts
1109 - Let Tests/datadesc_usage write the architecture on which the file was
1110 generated as first byte.
1111 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
1112 architecture descriptions.
1113 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
1114 on those architectures.
1115 - Optimization: send/recv array of scalar in one shoot
1117 2004-07-05 Martin Quinson
1118 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
1120 - Plug a whole bunch of memleaks
1121 - each process now have to call gras_{init,exit}. One day, their log
1122 settings will be separated
1123 - Continue the code factorisation between SG, RL and common in Transport.
1125 2004-07-04 Martin Quinson
1127 - Redistribution between SG and RL.
1128 We wanna have to accept in SG, so move accepted related parts of RL in
1129 the common part. (more precisely, the dynar of all known sockets is no
1130 more a static in transport.c, but part of the process_data)
1132 [gras_stub_generator]
1133 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
1135 2004-07-03 Martin Quinson
1136 - Create a new log channel tbx containing dict, set, log, dynar (to shut
1137 them all up in one shot)
1139 - Fix the ugly case of reference to dynamic array.
1140 - New (semi-public) function gras_datadesc_size to allow the messaging
1141 layer to malloc the needed space for the buffer.
1143 - gras_socket_close now expect the socket to close (and not its address to
1144 put NULL in it after it). This is because the socket passed to handlers
1145 is one of their argument (=> not writable).
1147 - propagate the interface cleanup from last week in datadesc, ie remove a
1148 superfluous level of indirection. User pass adress of variable
1149 containing data (both when sending and receiving), and not of a variable
1150 being a pointer to the data. Let's say that I like it better ;)
1151 The price for that is constructs like "int msg=*(int*)payload" in
1152 handlers, but it's a fine price, IMHO.
1154 - Let it work in RL (yuhu)
1156 2004-06-21 Martin Quinson
1158 - porting SG plugin and SG select to new standards (works almost).
1159 - plug memleaks and fix bugs around.
1162 - cleanup the prototype of data recv and force users to specify when they
1163 want to handle references to objects. Test case working even for cycles.
1164 - plug memleaks. Valgrind is perfectly ok with this.
1166 2004-06-12 Martin Quinson
1168 - cleanup the separation between plugin and main code in plugin creation
1170 2004-06-11 Martin Quinson
1172 - Reput hook for raw sockets, needed for BW experiments
1173 - kill a few lines of dead code
1174 [Data description] Interface cleanup
1175 - gras_datadesc_by_name returns the searched type or NULL.
1176 That way, no variable is needed to use a type desc once, which makes
1178 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
1179 two parameters were strings (field name, type name), leading to
1181 [Dicos] Interface cleanup
1182 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
1183 This is consistant with the dynar API.
1185 2004-04-21 Martin Quinson
1187 - Porting to new standards.
1189 - interface cleanup.
1190 There is no bag anymore, no need to take extra provision to mask the
1191 pointers behind "ID".
1192 Better splitup of functions between files create/exchange/convert.
1193 This is still a bit artificial since convert and receive are so
1194 interleaved, but anyway.
1196 - add a queued message list to procdata (the ones not matching criteria
1198 - factorize some more code between SG and RL wrt procdata
1200 - use gras_exit in example to track memleaks
1201 - get rid of gs_example now that GS is properly integrated into gras
1202 - update run_test to integrate the lastest tests (datadesc)
1204 - rename WARNINGn macros to WARNn since it prooved error-prone
1206 2004-04-19 Martin Quinson
1208 - register init/exit functions within gras module mechanism
1209 - send/receive function.
1210 Convertion is not implemented, but short-cutted if not needed.
1211 struct/array elements are sent one by one (instead of block-wise), but
1212 nobody really cares (yet). Get a prototype before optimizing.
1213 - tests (using a file socket) for DD send/receive on:
1214 - base types: int, float
1215 - array: fixed size, string (ie ref to dynamic string)
1216 - structure: homogeneous, heterogeneous
1217 - chained list, graph with cycle
1218 Believe it or not, valgrind is not too unhappy with the results. The
1219 cycle happily segfaults, but the others are ok. And I'm sick of pointers
1223 - Bugfix when using a filename explicitely (instead of '-')
1225 2004-04-09 Martin Quinson
1227 - factorize more code between RL and SG in socket creation
1228 - Complete the implementation and tests of:
1230 o file (only in RL, and mainly for debugging)
1232 I lost 3 days to design a portable address resolver, and then decided
1233 that the prototype mainly have to run on my box.
1234 Addressing portability too early may be like optimizing too early :-/
1236 - use gras_init in the Tests instead of the crappy parse_log_opt
1237 (the latter function is removed)
1238 [Conditional execution]
1239 - New functions: gras_if_RL/gras_if_SG (basic support for this)
1240 [Code reorganisation]
1241 - Get rid of libgrasutils.a since it makes more trouble than it solves.
1242 Build examples against the RL library, since there is no way to disable
1243 its creation for now.
1245 For information, the beginning of coding on GRAS was back in june
1246 2003. I guess that every line has been rewritten at least twice since