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
23 * Allow to control the simulation from a trace file.
24 New functions MSG_action_register() and MSG_action_trace_run()
25 The first one allows to associate a function execution to each
26 kind of action while the second one parses a trace file and
27 triggers the corresponding actions within the system.
28 For now, only a toy example is provided in examples/msg/actions
29 * Add an exemple of process migration in examples/msg/migration
30 * Fix a bug in task exchange which broke MSG_task_get_sender()
31 Add a teshsuite regression test for that.
32 [Bug: if MSG_task_get_sender() is called after sender exit,
36 * Add SIMIX_process_set_name() to change the name of the current
37 process in the log messages.
40 * Also include strbuff from xbt.h public header
41 * xbt_ex_display(): do not free the exception after displaying
42 This allows to do more with the given exception afterward.
43 Users should call xbt_ex_free() themselves.
44 * Do not allow malloc(0) since that's not portable
47 * Do not require doxygen in maintainer mode
49 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr>
51 SimGrid (3.3) stable; urgency=high
55 * JAVA BINDINGS for MSG (you dreamt of them? We made them)
58 * Introduce the SIMIX module: factorize code between MSG and GRAS.
61 Until now, GRAS were using MSG as an interface to SURF. It was
62 quite difficult because both interface have several differences
63 (MSG channels vs GRAS sockets were the most notable point).
65 This also opens the gate to SMPI (which should occur soon) and speed
66 up simulations by to 40% (even if it were not the main goal).
68 **************************************
69 *DO NOT MIX 3.2 RESULTS WITH 3.3 ONES* Simix may changes simulations!
70 **************************************
71 The point is that events occuring at the exact same timestamp are
72 not scheduled in the same order with the old and new version. This
73 may be enough to completely change the execution of simulations in
74 some cases. Sorry for the inconvenience.
76 * Cleanup and upgrade the XML format to push further scalability
77 issues (check http://hal.inria.fr/inria-00256883/ for more info)
79 * Improve the testing infrastructure with tesh. Now a very large part of
80 the code is tested not only by being run but also by checking that the
81 output match an expected output [Mt].
83 * Move on to FleXML v1.7 for the embeeded XML parsers. This version
84 is really less memory-demanding, which should allow you to use
85 larger files in SimGrid [AL].
87 * Inform valgrind about our contextes, so that it becomes usable
88 with the default (and more effecient) version of SimGrid
89 [contributed by Sékou Diakite, many thanks]
92 * Introduce a listener thread in charge of receiving incomming
93 messages from the network. It allows to overlap communication and
94 computation but most notably, it removes some stupid deadlocks due
95 to the fact that so far, a process could not send and receive at
96 the same time. This made most non trivial communication schema
98 * Convert the PIDs from long int to int to match the MSG ones (and
100 * New function: gras_agent_spawn() to launch a new process on
101 current host. Only working in simulation for now. [Mt]
102 * New function: gras_os_hostport() returning a constant form (ie,
103 not needing to be freed) of "gras_os_hostname():gras_os_myport()"
106 * Make the backtrace of exceptions more human readable [Mt]
107 * New module: xbt/str [Mt]
108 a ton of string utility functions (split, join, printf to a newly
109 allocated buffer, trim, etc)
110 * New module: xbt/hash [Mt]
111 SHA1 hashing algorithm (more to come if needed)
112 * New module: xbt/synchro [Mt]
113 synchronization tools (mutex and conditions) working the same way
114 in simulation and in real life (mainly useful for GRAS, but not
116 * New module: xbt/queue [Mt]
117 classical producer/consumer synchronization scheme
118 * xbt_dynar_new_sync() creates a synchronized dynar. All access
119 (using the classical functions will get serialized) [Mt]
120 * Make dictionary internal table dynamic. No need to specify its size
121 anymore; functions xbt_dict_new_ext() and xbt_dict_hashsize_set()
123 * Make sure the log channels are organized as a tree under windows
124 (because of ANSI C compatibility issue, any channel were child of
128 * Cleaned many thing in surf and fixed a few bugs [AL].
129 * Add a nice command line configuration mechanism to compose models [AL].
130 * Add a new model for parallel tasks (ptask_L07) that is less buggy than
131 the previous one (KCCFLN05). It relies on something that looks like
132 a max-min sharing mechanism but cannot be written as such. A new solver
133 was thus designed [AL].
134 * Add a new solver to lmm. Based on Lagrange optimization and
135 gradient-based descent, it enables to efficiently maximise systems s.a
137 sum f_i(x_i) s.t Ax<= b with A_{i,j}>=0 and f_i a concave function.
139 This solver enables to propose two new network models for TCP Reno and
140 TCP Vegas based on Low's work. These models still need to be fully
141 tested though [Pedro Velho].
144 * Bug fix in SD_simulate. Now the time bound given as argument is
146 * Use the new parallel task model (ptask_L07) as default.
147 * Use the SURF command line configuration mechanism.
148 * 0-size tasks (for synchronization) should now work.
150 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Sun Apr 12 05:20:36 CEST 2009
152 SimGrid (3.2) stable; urgency=high
156 We still experience issues on this platform, but we believe that at
159 GRAS API BREAKAGE (for simplification purpose, sorry):
160 * the gras_msgtype_by_name is not used anymore. Instead of
161 gras_msg_send(toserver, gras_msgtype_by_name("request"), &request);
162 you can write (and must)
163 gras_msg_send(toserver, "request", &request);
164 - If you still want to pass a gras_msgtype_t to the function (to cache
165 the type and avoid the lookup time), use the gras_msg_send_() variant.
166 - Impacted functions:
167 gras_cb_register, gras_cb_unregister, gras_msg_send, gras_msg_wait,
168 gras_msg_rpccall, gras_msg_rpc_async_call, gras_msg_wait_ext
169 * The callbacks are now expected to return 0 when everything went well
170 (just like the main() function)
172 GRAS new features and improvements:
173 * New module mecanism where user code can use per process globals [Mt]
174 This is similar to gras_userdata_*() functions, but for libraries. It
175 factorize some code developped over and over in the examples and AMOK.
176 It has still to be documented and used (only amok/peermanagement is
178 * Fix a vicious bug in the TCP buffering mecanism which leaded to message
179 loss when they were small enough to fit into the buffer and sent quickly
180 enough so that they can all get received in one shoot.
181 * gras_datadesc_by_name and gras_msgtype_by_name: now raise an exception
182 if not found. Use the *_or_null() variant for the old semantic.
183 * In gras_msg_handle, do not discard messages without callback.
184 They are probably messages to be explicitly awaited later (ie, proofs of
185 mis-synchronization in userland since they are sent before being awaited)
187 * gras_socket_meas_send/recv: semantic changed!
188 The numerical arguments used to be (1) the total amount of data to send
189 and (2) msg_size. This was changed to (1) msg_size and (2) amount of
190 messages. This was need for the fool willing to send more than MAXINT
191 bytes on quite fat pipes.
194 * Do really rename the hostmanagement module to peermanagement. [Mt]
195 Ie, rename functions from amok_hm_* to amok_pm_*. This breaks the API,
196 but this is rather new and this was documented in the module
197 documentation (poor excuses, I admit)
198 * Bandwidth measurement semantic changed! This follows the changes to
199 gras_socket_meas_send/recv explained above.
202 * A sequential mode has been added to the workstations. When a workstation
203 is in sequential mode, it can execute only one task, and the other tasks
204 are waiting in a FIFO. [Christophe Thiery]
207 * The KCCFLN05 workstation model now handles parallel tasks. It is the
208 model for SIMDAG. [Christophe Thiery]
209 * Bug fix in the maxmin solver: Some values were close to 0 instead of
210 equal to 0, which caused some bad behaviors in
211 saturated_constraint_set_update. I now use a threshold mechanism like in
215 * When running manually src/testall, you select specific units [Mt]
216 testall is the result of our cunit mecanism, and should replace all
217 the scripty thingy around since bash don't run easily on billware.
219 * A mallocator system has been added. [Christophe Thiery]
220 Mallocators allow you to recycle your unused objects instead of freeing them
221 and allocating new ones.
223 Documentation update:
224 * FAQ reworking + New FAQs:
225 - "Valgrind spits tons of errors!" [Mt]
226 - "How to repport bugs" [Mt]
227 - "Cross-compiling a Windows DLL of SimGrid from Linux" [Mt]
228 - "What is the difference between MSG, SimDag, and GRAS?" [Mt]
229 - Communication time measurement within MSG [AL]
230 - I experience weird communication times when I change the latency [AL]
233 - an introduction to the framework and to the used communication model
234 - an initiatic tour introducing the most proheminent features:
236 . Lesson 0: Installing GRAS
237 . Lesson 1: Setting up your own project
238 o Part 2: Message passing
239 . Lesson 2: Exchanging simple messages
240 . Lesson 3: Passing arguments to the processes (in SG)
241 . Lesson 4: Attaching callbacks to messages
242 . Lesson 5: Using globals in processes
243 . Lesson 6: Logging informations properly
244 . Lesson 7: Using internal timers
245 . Lesson 8: Handling errors through exceptions
246 . Lesson 9: Exchanging simple data
247 . Lesson 10: Remote Procedure Calling (RPC)
248 . Lesson 11: Explicitely waiting for messages
249 . Recapping of message passing features in GRAS
250 - A HOWTO section containing:
251 o HOWTO design a GRAS application
252 More are due, of course. They will come latter. In the meanwhile, you can
253 check the examples which are still here.
255 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri Mar 16 21:11:46 CET 2007
257 SimGrid (3.1) stable; urgency=high
261 There was a stack corruption somewhere, visible only when optimizing
262 with these versions. [Vince]
265 * This is a NEW module! SimDAG (SD for short) is a revival of the old SG
266 module that enabled to play with Directed Acyclic Graphs. It is built
267 directly on top of SURF and provides an API rather close to the old
268 SG. Some old codes using SG are currently under rewrite to check that
269 all needful functions are provided. [Christophe Thiery]
272 * Complete rewrite of the KCCFLN05 workstation model. It is now an
273 extension of the classical CLM03 model that gracefully handles
274 failures. This is now the default model for MSG and GRAS. It doesn't
275 handle parallel tasks yet though. [AL]
276 * Bug fix: Weights were not correctly set in the network part.
277 WARNING: This may have resulted in incorrect results with simulations
278 where there are more than one flow on a given link. [AL]
281 * After a (long ?) discussion on simgrid-devel, we have decided that the
282 convention we had on units was stupid. That is why it has been decided
283 to move from (MBits, MFlops, seconds) to (Bits, Flops, seconds).
284 WARNING : This means that all previous platform files will not work as
285 such with this version! A warning is issued to ask users to update
287 A conversion script can be found in the contrib module of the CVS, under
288 the name contrib/platform_generation/surfxml_update.pl [MQ]
291 * Bug fix: Processes were started in reverse order, wrt deployment file.
292 WARNING: if your code relies on this bug, please fix it. [AL]
293 * Bug fix: Add a test in MSG_task_execute to stop whenever a task is
294 being executed on two different locations. [AL]
295 * Bug fix: Failures are now better supported thanks to Derrick's tests
296 (there was many failure situations I hadn't thought of and that weren't
297 correctly handled). [AL]
298 * New function: MSG_host_is_avail indicates you whether a given m_host_t
302 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
303 Exception occurring on server-side are propagated back to client (!).
305 API CHANGE: the callback changed their prototype. Change:
306 int my_handler(gras_socket_t expeditor, void *payload_data) {
308 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
309 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
311 * New! function: gras_msg_handleall to deal with all messages arriving
312 within a given period.
313 * New! function: gras_socket_server_range to get a server socket in a
314 range of port numbers (ease to avoid port number conflicts) [MQ]
315 * New! gras processes display their backtrace when they get a SIGUSR1
316 or when Ctrl-C is pressed. Use Ctrl-C Ctrl-C to exit.
317 Sweet to debug RL processes [MQ]
321 - Do not force experiment sizes to be expressed in kb, or it becomes
322 impossible to measure the latency this way (needs one byte-long tests)
323 WARNING: this changes the amok_bw_* function semantic. [MQ]
324 - Implements the link saturation stuff. [MQ]
325 * Peer management module:
326 New! module factorizing code that we wrote over and over [MQ].
329 * New module: cunit (my jUnit implementation in ansi C) [MQ]
330 - Test units are placed directly into the library code, they get extracted
331 automatically and placed into the src/testall binary.
332 - Convert most of the XBT tests to this system.
333 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
334 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
335 live in the same C code. This required to change a little bit the API
336 of surfxml parsing but shouldn't be an issue for end-users. [AL]
337 * New module: sparse graph structure with basic algorithms (this is work
338 in progress and the API is not considered to be frozen yet). [AL]
339 * Display more information on backtraces: source line & function names are
340 now displayed just like valgrind does (rely on addr2line tool) [MQ]
341 * New function: xbt_backtrace_display(). Sweet while debuging [MQ]
342 * Reworked a little bit some #include statements to load only required
343 headers. Some user code that relied on SimGrid to include stdlib or
344 stdio may need to include it by themselves. [AL]
345 * Fixed xbt/log.h. A missing SG_BEGIN_DECL prevented compilation with
347 * Renamed xbt_host_t into xbt_peer_t since it betterly describes what I
348 meant. This breaks the API of AMOK and of xbt/config. Sorry about this,
349 but I guess that almost nobody used those parts. [MQ]
351 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 14 Jul 2006 01:32:27 +0200
353 SimGrid (3.0.1) stable; urgency=low
356 * Unfortunately, I had missed 5 misnamed functions:
357 xbt_fifo_item_t xbt_fifo_newitem(void);
358 void xbt_fifo_freeitem(xbt_fifo_item_t);
359 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
360 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
361 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
362 They're now deprecated. Please use their new versions:
363 xbt_fifo_item_t xbt_fifo_new_item(void);
364 void xbt_fifo_free_item(xbt_fifo_item_t);
365 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
366 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
367 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
369 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
370 * Documentation: xbt_log module unmercifully reworked [MQ]
371 * Bugfix: there was a problem with the ending of contexts with
372 the pthread backend. It caused some weird deadlock or behavior
373 depending on the pthread implementation. [AL]
374 * Bugfix: get the exceptions raised in the simulator repport where
375 and why they come from when they are not catched in time [AL, MQ]
378 * Bugfix: Do repport the error when two non-connected hosts try to
379 exchange data (Thanks to Flavien for stumbling into this one) [AL]
382 * Add additionnal checkings on communications. Assert that two
383 communicating hosts are connected by a set of links... [AL]
386 * Add additionnal checkings on channel values in communication [AL]
387 * New: MSG_task_get_source to see on which host a task was generated [HC]
388 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
389 the number of tasks waiting to be received on channel and sent
391 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
392 waits for the first task coming from a given host.. [AL]
394 GRAS new functionnalities: [MQ]
395 * Enhance the parsing macro to allow the size of multidimentional objects
396 to be given thru annotations.
397 * New example (and documentation): Matrix Multiplication a la RPC
398 (as when I was young!) and fix a bunch of bugs found on the way.
400 GRAS performance improvements: [MQ]
402 * Reduce the amount of cbps creation/destruction by making it static to
403 datadesc_send/recv() and using a (newly created) cbps_reset (based on
406 * Change libdata to a set so that we can search for stuff by ID (and thus
407 reduce the insane amount of dict lookups)
410 * Actually implement gras_datadesc_copy() so that we don't have to mimick
411 RL communication on top of SG since it's so uneffective.
412 It may also be used for inter-thread communication in RL, one day.
413 * Use gras_datadesc_copy() to exchange messages on top of SG
415 - improve message exchange performance on top of SG
416 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
417 * Don't exchange on the network the size of the used part of buffer,
418 instead, specify the possible buffer size to read().
420 - reduces the amount of read/write calls (one pair per exchange)
421 - reduces the amount of exchanged data (the size)
422 - allows to retrieve all arrived data on receiver side, if we don't need
423 it right now (subsequent read will peek the buffer)
424 - allows the receiver to proceed with the begining of the stream before
425 everything is arrived
426 - make it possible to build an iov transport (using readv/writev)
428 - take care of the data with non-stable storage (like stacked data),
430 * If possible, TCP send uses vector I/O (when writev() is here)
431 - Don't use it for receive since we send data sizes and data on the
432 same stream, so we wouldn't be able to chain large amount of chunks
433 before having to flush the stuff to read the size.
434 * Rework the transport plugin mecanism to simplify it and reduce the
435 amount of pointer dereferencement when searching for the right function
438 * I guess that now, we do almost as few system calls as possible while
439 doing as few data copy as possible.
441 To improve it further, we could try to send all the sizes first and then
442 all the data (to use iov on receiving size), but it's only a partial
443 solution: when you have 2 dimensional data, the sizes of the second
444 dimension is data of the first dimension, so you need 3 streams.
446 I'm not sure the potential performance gains justify the coding burden.
448 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
450 SimGrid (3.00) stable; urgency=high
453 * New! Give the possibility to hijack the surf parser and thus bypass
454 MSG_create_environment and MSG_launch_application. Have a look at
455 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
457 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
459 SimGrid (2.96) unstable; urgency=low
464 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
465 This deprecates the xbt_error_t mecanisms.
466 It modifies (simplifies) all XBT and GRAS API.
467 MSG API keeps unchanged (exceptions raised by XBT are catched from
468 within MSG and masked with existing error handling facilities)
471 * New! Add a FATPIPE model. [AL]
472 * New! Add a parallel task model. [AL]
473 * New! Add automatically a loopback interface (in the default
474 network model) if none was precised.
477 * Bugfix: MSG_process_resume now works with the current running process.
479 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
480 * Modification of MSG_task_get_compute_duration. Once a task has been
481 processed, the value returned by this function is now equal to 0. [AL]
482 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
483 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
484 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
485 of a m_task. For now, MSG_task_cancel only works with computation
487 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
488 of the processor (in Mflop/s) regardless of the current load on the
490 * API Change: use proper naming convention for MSG_getClock and
491 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
493 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
494 This function changes the priority of a computation task. This priority
495 doesn't affect the transfer rate. A priority of 2 will make a task
496 receive two times more cpu power than the other ones. This function
497 has been added to suit the needs of Nguyen The Loc and hasn't been that
498 much tested yet. So if it fails, please report it and send me your code.
500 * API Change: removed all functions and types that were marked "deprecated"
501 since many months. Renamed MSG_global_init_args to MSG_global_init.
503 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
505 SimGrid (2.95) unstable; urgency=low
508 * Steal some nice code to GNU pth to fix context detection and usage [AL]
509 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
510 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
513 * Bugfix: Allow absolute paths for platform description files [MQ]
514 * Bugfix: do free the variables after use. Leads to drastic performance
516 * Implement max_duration (ie, timeouts) on resources [AL]
519 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
521 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
522 the needed support (timeouts) [AL]
525 * Implement measurement sockets. You can now get the bandwidth between two
526 hosts thanks to AMOK (see below). [MQ]
527 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
528 dynar over the network (yeah) [MQ]
529 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
530 * simple/static token-ring example. [Alexandre Colucci and MQ]
531 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
535 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
536 type parsing. "short" and "long" modifiers were also ignored; other
537 modifier (such as reference level) are still ignored. [MQ]
538 * Embeed the buffer size within the buffer itself on SG. [MQ]
539 That way, send() are atomic and cannot get intermixed anymore (at least
540 the ones which are less than 100k; bigger messages still have the issue)
541 * Array size pushed by the field, not by the field type (or each
542 and every long int will push stuff to the cbps) [MQ]
543 * use select() to sleep since it allows to portably sleep less than one
546 GRAS (minor cleanups)
547 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
550 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
551 * type category 'ignored' killed as it was never used and were difficult
553 * whether a type can cycle or not is now a flag, leaving room for more
554 flags (as "ignored", if we feel the need one day ;)
555 * Rename raw sockets to measurement sockets since "raw" has another
556 meaning in networking community.
559 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
560 and offers features not belonging to GRAS but that most applications
561 need. One day, it may be a set of plugins loadable at runtime.
562 * New module: bandwidth
563 bandwidth measurement between arbitrary nodes running this module. [MQ]
565 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
567 SimGrid (2.94) unstable; urgency=low
569 The first beta release of SimGrid 3 !
573 * Update the main page and the FAQ. Adding references to gforge.
576 * Add a gras_os_getpid function.
579 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
580 * Extend the logs so that they also print PID, hostname, date, ... if
582 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
583 and kill the old version which were in testsuite/
584 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
588 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
589 for users. Instead, each programming environment (GRAS, MSG,...) use it
590 when needed to provide such a feature to users.
591 Don't shortcut the mecanism or you will also shortcut the virtualization
592 you need on the simulator.
596 * Cleanups in configury with regard to compile optimization/warning flags.
597 Also add -fno-loop-optimize to any powerpc since it's the optimization
598 killing gcc (< 3.4.0).
599 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
600 needed by MSG examples complications
601 * Borrow configury beautifications from PHP
604 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
605 warning (thanks loris for stumbling into this one).
606 * Bugfix: stop loading private headers (gras_config.h) from the public
610 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
611 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
612 RL would imply a DNS resolver, which is *hard* to do in a portable way
613 (and therefore delayed).
614 * Implement a real timer mecanism and use it in timing macros. This allows
615 to avoid rounding errors and get a 0.000005 sec precision in timing
616 macros. While I was at it, various cleanups:
617 - allow to declare more than one timed section per file (fix a stupid bug)
618 - move some private declaration to the right place
619 - merge conditional execution and timing macros into emulation module
620 - document the module
621 - make sure the module cleanups its mess on gras_exit
622 * Documentation improvements:
623 - (new) how to compile applications using GRAS
624 - (new) emulation support (timing macros)
626 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
628 SimGrid (2.93) unstable; urgency=low
630 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
633 - Use Paje properly where used. Still to be sanitized properly.
634 - Portability fix: Add an implementation of the contexts using pthread
638 - Add xbt_procname(): returns the name of the current process.
639 Use it to show the current process's name in all logging.
641 - fix detection of older flex version and the reaction, since we do
642 depend on modern ones (we use lex_destroy)
643 - Better separation of SG and RL in the libs: remove all simulation code
644 from libgras. As a result, this lib is now only 200k when stripped.
645 Some of the xbt modules may also be duplicated (two sets and such) and
646 should be cleaned/killed before SG3.
647 - Insist on using xlC on AIX because of weird problems involving gcc there.
648 - Cleanup the make remote stuff. This is now done by scripts
649 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
650 mainly for our private use, but we're working on changing them to user
653 - Bugfix: flush the socket on close only if there is some *output*.
654 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
656 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
659 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
661 SimGrid (2.92) unstable; urgency=low
663 Alpha 3 on the path to SimGrid 3
667 - New! First try of benchmarking macros.
668 - New! First try so that gras_stub_generator generate deployment and
669 remote compilation helpers.
671 - Bugfix: Initialization fix in msg_test.
675 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
677 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
679 - API Change: gras_os_sleep to take the amount of seconds as a double.
680 Accepting an int was error prone since it was the only location where
681 seconds were coded as such. It leaded to damn rounding errors.
682 - Bugfix: Hard to belive that timers ever worked before this.
684 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
686 SimGrid (2.91) unstable; urgency=low
688 Alpha 2 on the path to SimGrid 3
692 - Bug fix in the lmm_solver.
694 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
695 through the function MSG_paje_output.
696 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
697 - It is possible to bound the rate of a communication in MSG with
698 MSG_task_put_bounded() (was already in the previous version but I had forgotten
699 to write it in the changelog).
700 - Bug fix to let GRAS run on top of MSG until we move it directly on top
705 - Various cleanups to the autotools stuff
706 - Begin to move Gras examples to examples/gras/
707 - Let make distcheck work again (yeah!)
709 - documentation overhauled using doxygen.
710 gtk-doc-tools is dead in SimGrid now.
711 - Automatically extract all existing logging categories, and add the list
712 to the documentation (long standing one, to say the less)
714 - Cleanup the known architecture table. Reorder the entries to group what
715 should be, and use a more consistent naming scheme.
716 (some of the test dataset are still to be regenerated)
717 - New! Allow library to register globals on each process just as userdata
719 This is implemented using a xbt_dict and not a xbt_set, so we loose the
720 lookup time (for now).
721 Use it in msg and trp.
722 This cleans a lot the internals and helps enforcing privacy of the
723 headers between the gras components.
724 - New! Add a timer mechanism, not unlike cron(8) and at(1).
725 - Bugfix: gras_os_time was delirious in RL.
726 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
728 - Reenable GRAS now that it works.
730 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
732 SimGrid (2.90) unstable; urgency=low
734 Alpha 1 on the path to SimGrid 3
736 * It is a long time since the last release of SimGrid. I'm sorry about
737 that but as I had told you, I was rewriting a lot of things. I apologize
738 to those who had been reporting bugs to me and that I had not answered.
739 If your bug is still in the new version, please tell me. Here is a
740 summary of the main changes.
742 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
743 lead by Martin Quinson. As a consequence SimGrid gains a lot in
744 portability, speed, and a lot more but you'll figure it out later.
745 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
746 I wanted to release the new MSG as soon as possible and I have
747 broken GRAS, which is the reason why, for now, only MSG is fully
748 functional. A laconic description of these projects is available
749 in the documentation.
751 * REVOLUTION 2: I have removed SG and I am now using a new simulation
752 kernel optimized for our needs (called SURF but only the developers
753 should use it). Hence, MSG is now roughly 30 times faster and I think
754 that by rewriting a little bit MSG, I could event speed it up a little
755 bit more. Beside the gain in speed, it is also much easier to encode a
756 new platform model with SURF than it was with SG. More to come...
758 * REVOLUTION 3: I have tried to change a little as possible the API of
759 MSG but a few things really had to disappear. The main differences
760 with the previous version are :
761 1) no more m_links_t and the corresponding functions. Platforms are
762 directly read from a XML description and cannot be hard-coded
763 anymore. The same format is used for application deployment
764 description. The new format is described in the documentation.
765 Have a look in tools/platform_generation. There is a tiny script
766 that converts from the old platform format to the new one. Concerning
767 the application deployment format, parsing the old one is tricky.
768 I think most of you should however be able to convert your files. If
769 it is really an issue, I can write a C code that does the conversion.
771 2) the toolbox tbx does not exist anymore. We now have a library
772 with much more data-structures but without the hash-tables (we have
773 dictionaries that are much faster).
775 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
777 *****************************************************************************
778 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
779 * don't feel like converting the oldies. (Mt) *
780 *****************************************************************************
783 Version 2.90: "the long awaited one"
784 - Finished rewriting and debugging MSG. Rewrote the documentation.
785 - disable GRAS for now since it needs to be ported to the newest SG
788 - Finish the port to windows (using mingw32 for cross-compile)
791 - Main loop and datastructures of SURF. A cpu resource object is
792 functional. Surf can thus be used to create cpu's with variable
793 performance on which you can execute some actions.
795 2004-11-15 Martin Quinson
796 - Port to ARM. Simply added the alignment and size descriptions. Should
797 work, but the ARM machines are so slow that I didn't had the opportunity
798 to 'make check' over there yet.
800 2004-11-15 Arnaud Legrand
801 - Trace manager now written. It uses a heap structure and is therefore
802 expected to be efficient. It may however be speeded up (particularly
803 when many events occur at the same date) by using red and black
804 trees. One day maybe...
805 - Max-min linear system solver written. It uses a sparse matrix
806 structure taking advantage of its expected use. Most operations are
807 O(1) and free/calloc are called as few as possible. The computation of
808 the minimum could however be improved by using a red and black tree
811 2004-11-03 Arnaud Legrand
812 - Rename every gras_* function that was in xbt/ to its xbt_
814 - Add a heap and a doubly-linked list to xbt
815 - Added a dichotomy to the dictionaries. make check works as well before
816 so I assume that the patch is correct. I do not know however if things
817 run effectively faster than before now. :)
819 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
820 SimGrid, and the version number is bumped to 2.x
822 2004-10-29 Martin Quinson
823 - Introduction of the remote errors.
824 They are the result of a RMI/RPC on the remote machine.
825 ErrCodes being scalar values, you can't get the host on which those
826 errors did happen. Extending the error mechanism as in Gnome is possible.
827 No idea yet whether it is a good idea.
829 2004-10-28 Martin Quinson
830 - Interface revolution: the Starred Structure Eradication.
831 I used to do typedef struct {} toto_t; and then handle *toto_t.
832 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
833 - ???_t is a valid type (builded with typedef)
834 - s_toto_t is a structure (access to fields with .)
835 - s_toto is a structure needing 'struct' keyword to be used
836 - e_toto_t is an enum
837 - toto_t is an 'object' (struct*)
839 typedef struct s_toto {} s_toto_t, *toto_t;
840 typedef enum {} e_toto_t;
841 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
842 s_toto_t) is private.
844 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
845 so that it can set the variable to NULL. It was so for dicts and sets,
846 it changed for dynars.
848 - Fix a bunch of memleaks in dict_remove
849 - Fix a bug in sg/server_socket opening: it failed all the time.
851 2004-10-07 Martin Quinson
852 - Speed up dynar lookup operation a bit.
854 gras_dynar_get is dead.
856 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
857 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
858 (which gives you the address of the stored data).
860 gras_dynar_get_as is an helpful macro which allows you to retrieve a
861 copy of the data using an affectation to do the job and not a memcpy.
863 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
864 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
866 It does not really speedup the dynar test because they are
867 setting elements all the time (and look them seldom). But the dict does
868 far more lookup than setting.
870 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
872 2004-10-05 Martin Quinson
873 - Allow to (en/dis)able the cycle detection at run time.
875 Whether we should check for cycle or not is now a property of each
876 datatype. When you think there may be some cycle, use datadesc_cycle_set.
877 datadesc_cycle_unset allow to remove this property when previously set.
879 Note that the cycle detection is off by default since it impacts the
880 performance. Watch the data you feed GRAS with ;)
882 This property is hereditary. Any element embedded in a structure having it
883 set have it set for the time of this data exchange.
885 You should set it both on sender and receiver side. If you don't set it on
886 sender side, it will enter an endless loop. If you forget on receiver
887 side, the cycles won't be recreated after communication.
889 - Header reorganization.
890 Kill gras_private.h, each submodule must load the headers it needs.
892 2004-10-04 Martin Quinson
893 - Interface revolution: do not try to survive to malloc failure.
895 Now, gras_malloc and friends call gras_abort() on failure.
896 As a conclusion, malloc_error is not a valid error anymore, and all
897 functions for which it was the only gras_error_t return value are
898 changed. They now return void, or there result directly.
899 This simplify the API a lot.
901 2004-09-29 Martin Quinson
902 - Re-enable raw sockets.
903 Created by gras_socket_{client,server}_ext;
904 Used with gras_raw_{send,recv}
907 It should allow to kill the last bits of gras first version soon.
909 This is not completely satisfactory yet (duplicate code with
910 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
913 - Simplify transport plugin (internal) interface by not passing any
914 argument to _server and _client, but embedding them in the socket
917 2004-09-28 Martin Quinson
918 - Finish the port to AIX.
919 autoconf was my problem (segfault within the malloc replacement
920 function. No idea why)
922 2004-09-16 Martin Quinson
923 - Fix some size_t madness on 64bit architectures.
925 2004-09-08 Martin Quinson
926 - Reduce the number of system headers loaded, overload some more system
927 calls (such as malloc to cast the result of the system one, and work
929 - Fix and reintroduce the config support
931 2004-09-07 Martin Quinson
932 - Source code reorganization to allow Arnaud to surf all over there.
933 - Allow to document the logging categories.
934 - Remove all uppercase from logging categories and useless cleanup in names.
936 2004-08-18 Martin Quinson
937 Version 0.6.2 (protocol not changed; API changed)
938 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
939 gras_error_t), and NULL when not found. Functions expecting a msgtype
940 as argument (msg_wait; msg_send) deal with NULL argument by providing a
941 hopefully usefull message.
942 - Portability to prehistoric sparcs again
944 2004-08-17 Martin Quinson
945 Version 0.6.1 (protocol not changed; ABI not changed)
946 - prealloc some buffers to speed things up
948 2004-08-11 Martin Quinson
949 Version 0.6 (protocol not changed; ABI expended)
950 - The parsing macro can deal with the references, provided that you add
951 the relevant annotations (using GRAS_ANNOTE(size,field_name))
953 2004-08-09 Martin Quinson
954 Version 0.5 (protocol not changed; ABI changed)
955 - Allow to off turn the cycle detection code in data exchange at
956 compilation time. It should be at run time, but I'm short of time (and
957 the config stuff is still broken). That way, we keep dict out of the
958 critical path, which is good because the performance is poor:
959 - search not dichotomial yet
960 - dynar give no way to access their content and memcpy everytime
961 - In composed data description (struct, ref and so on), stop foolness of
962 keeping the subtype's ID, but store the type itself. This keeps sets out
963 of the critical path, which is good since they rely on dynar and
964 dictionnaries. The only loose of that is that we cannot detect the
965 redeclaration of a structure/union with another content (but I'm not sure
966 the code detected well this error before anyway). We still can detect
967 the redefinition discrepancy for the other types.
968 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
969 it breaks the code because of type-punning used all over the place).
970 This breaks on all non-gcc architectures (for now).
972 All those changes (plus the buffer of last time) allow me to gain 2 order
973 of magnitude on cruel tests consisting of 800000 array of integers on two
974 level of a hierarchical structure (200 secondes -> 4 secondes)
977 - the selector of reference must now return the type it points to, not
980 2004-08-06 Martin Quinson
981 Version 0.4 (protocol changed; ABI not changed)
982 - Allow to pass --gras-log argument to processes in simulation mode. Really.
983 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
984 - Implement a buffer transport, and use it by default (it relies on tcp in
985 real life and on sg in simulation).
986 That's a bit hackish since I had a new field to the structure to store
987 its data without interfering with the subtype ones. Inheritance
988 is tricky in C. And that's a kind of reverse inheritance with one class
989 derivating two classes. Or maybe a game with java interfaces. Anyway,
990 that's damn hard in C (at least).
991 Moreover, I got tired while trying to ensure plugin separation and
992 genericity in SG mode. MSG wants me to do weird things, so let's go for
993 cruel hacks (temporarily of course ;).
994 See comment in transport_private.h:71
995 - do not include all the _interface headers in private but in the files
996 which really need them (to cut the compilation time when they are
999 2004-07-26 Martin Quinson
1000 Version 0.3 (protocol not changed; ABI changed)
1001 - Major overhault of the datadesc interface to simplify it:
1002 - shorted the function names:
1003 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
1004 - add a trivial way to push/pop integers into the cbps without malloc.
1005 This allows to make really generic sub_type description, which simply
1006 pop their size of the stack.
1007 - add a function gras_datadesc_ref_pop_arr() which does what users want
1008 most of the time: Declare a dynamic array (which pops its size of the
1009 stack) and declare a reference to it. Poor name, but anyway.
1010 - kill the post-send callback, add a post-receive one
1012 2004-07-23 Martin Quinson
1013 Version 0.2 (protocol changed; ABI changed)
1014 - add some testing for cpbs in the test cases, and fix some more bugs.
1015 This invalidate again the little64 data file, since I cannot regenerate
1017 - remove an awfull optimization in the logging stuff, allowing me to:
1018 - understand it again
1019 - learn gcc how to check that the argument match the provided format
1020 - fix all errors revealed by gcc after that
1021 - internal keys of dict are not \0 terminated. Deal with it properly in
1022 loggings instead of segfaulting when the user want to see the logs :-/
1024 2004-07-22 Martin Quinson
1025 - Fix some stupid bug preventing cbps (callback postit) from working
1027 2004-07-21 Martin Quinson
1028 - Some documentation cleanups
1029 - remove the useless last argument of msgtype_declare
1030 - rename the Virtu functions to fit into the 'os' namespace
1031 - move headers src/include -> src/include/gras/ and stop fooling with
1032 gras -> . symbolic link
1033 - make distcheck is now successful
1035 2004-07-19 Martin Quinson
1037 - Build shared library also
1038 - Install html doc to the right location
1039 - stop removing maintainer files in make clean
1040 - build tests only on make check
1042 2004-07-13 Martin Quinson
1044 - No major issue in previous version => change versionning schema
1045 - Re-enable little64 convertion test now that Abdou kindly regenerated the
1046 corresponding dataset.
1048 2004-07-11 Martin Quinson
1050 - Get it working with any kind of structure (we can compute the padding
1051 bytes remotely for all the architectures I have access to)
1052 - Implement the structure parsing macro (still not quite robust/complete)
1053 - Improvement to the remote testing toysuite
1055 2004-07-10 Martin Quinson
1056 [autoconf mechanism]
1057 - get ride of a bunch of deprecated macros
1058 - actually run the test for two-compliment, not only compile it :-/
1059 - test whether the structures get packed (and bail out if yes. Damn.
1060 Alignment is a serious matter)
1061 - test whether the structures get compacted (but respecting the alignment
1062 constraints of each types)
1063 - test whether the array fields of structures can straddle alignment boundaries
1065 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
1066 (took me 2 hours to find that bug, looking at the wrong place)
1067 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
1068 before sending/receiving and are used to compute the offsets of fields
1069 - Given that padding size depend even on compiler options, keep track of
1070 alignment and aligned_size only for the current architecture. Not a big
1071 deal since we send structure fields one after the other (seems
1073 - Add the datastructure used for IEEE paper by the PBIO guys to the test
1074 program, let it work on linux/gcc/little32. portability todo.
1076 2004-07-08 Martin Quinson
1077 - import and improve remote compilation support from FAST
1078 - make sure make check works on half a dozen of machines out there
1080 2004-07-07 Martin Quinson
1081 Let's say it's version 0.0.3 ;)
1082 - Implement conversions (yuhu!)
1083 - Let it work on solaris (beside conversion, of course)
1084 - Stupid me, using rand() to generate the conversion datatests in not wise.
1086 2004-07-06 Martin Quinson
1087 - Let make dist work, since I'm gonna need it to compile on remote hosts
1088 - Let Tests/datadesc_usage write the architecture on which the file was
1089 generated as first byte.
1090 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
1091 architecture descriptions.
1092 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
1093 on those architectures.
1094 - Optimization: send/recv array of scalar in one shoot
1096 2004-07-05 Martin Quinson
1097 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
1099 - Plug a whole bunch of memleaks
1100 - each process now have to call gras_{init,exit}. One day, their log
1101 settings will be separated
1102 - Continue the code factorisation between SG, RL and common in Transport.
1104 2004-07-04 Martin Quinson
1106 - Redistribution between SG and RL.
1107 We wanna have to accept in SG, so move accepted related parts of RL in
1108 the common part. (more precisely, the dynar of all known sockets is no
1109 more a static in transport.c, but part of the process_data)
1111 [gras_stub_generator]
1112 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
1114 2004-07-03 Martin Quinson
1115 - Create a new log channel tbx containing dict, set, log, dynar (to shut
1116 them all up in one shot)
1118 - Fix the ugly case of reference to dynamic array.
1119 - New (semi-public) function gras_datadesc_size to allow the messaging
1120 layer to malloc the needed space for the buffer.
1122 - gras_socket_close now expect the socket to close (and not its address to
1123 put NULL in it after it). This is because the socket passed to handlers
1124 is one of their argument (=> not writable).
1126 - propagate the interface cleanup from last week in datadesc, ie remove a
1127 superfluous level of indirection. User pass adress of variable
1128 containing data (both when sending and receiving), and not of a variable
1129 being a pointer to the data. Let's say that I like it better ;)
1130 The price for that is constructs like "int msg=*(int*)payload" in
1131 handlers, but it's a fine price, IMHO.
1133 - Let it work in RL (yuhu)
1135 2004-06-21 Martin Quinson
1137 - porting SG plugin and SG select to new standards (works almost).
1138 - plug memleaks and fix bugs around.
1141 - cleanup the prototype of data recv and force users to specify when they
1142 want to handle references to objects. Test case working even for cycles.
1143 - plug memleaks. Valgrind is perfectly ok with this.
1145 2004-06-12 Martin Quinson
1147 - cleanup the separation between plugin and main code in plugin creation
1149 2004-06-11 Martin Quinson
1151 - Reput hook for raw sockets, needed for BW experiments
1152 - kill a few lines of dead code
1153 [Data description] Interface cleanup
1154 - gras_datadesc_by_name returns the searched type or NULL.
1155 That way, no variable is needed to use a type desc once, which makes
1157 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
1158 two parameters were strings (field name, type name), leading to
1160 [Dicos] Interface cleanup
1161 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
1162 This is consistant with the dynar API.
1164 2004-04-21 Martin Quinson
1166 - Porting to new standards.
1168 - interface cleanup.
1169 There is no bag anymore, no need to take extra provision to mask the
1170 pointers behind "ID".
1171 Better splitup of functions between files create/exchange/convert.
1172 This is still a bit artificial since convert and receive are so
1173 interleaved, but anyway.
1175 - add a queued message list to procdata (the ones not matching criteria
1177 - factorize some more code between SG and RL wrt procdata
1179 - use gras_exit in example to track memleaks
1180 - get rid of gs_example now that GS is properly integrated into gras
1181 - update run_test to integrate the lastest tests (datadesc)
1183 - rename WARNINGn macros to WARNn since it prooved error-prone
1185 2004-04-19 Martin Quinson
1187 - register init/exit functions within gras module mechanism
1188 - send/receive function.
1189 Convertion is not implemented, but short-cutted if not needed.
1190 struct/array elements are sent one by one (instead of block-wise), but
1191 nobody really cares (yet). Get a prototype before optimizing.
1192 - tests (using a file socket) for DD send/receive on:
1193 - base types: int, float
1194 - array: fixed size, string (ie ref to dynamic string)
1195 - structure: homogeneous, heterogeneous
1196 - chained list, graph with cycle
1197 Believe it or not, valgrind is not too unhappy with the results. The
1198 cycle happily segfaults, but the others are ok. And I'm sick of pointers
1202 - Bugfix when using a filename explicitely (instead of '-')
1204 2004-04-09 Martin Quinson
1206 - factorize more code between RL and SG in socket creation
1207 - Complete the implementation and tests of:
1209 o file (only in RL, and mainly for debugging)
1211 I lost 3 days to design a portable address resolver, and then decided
1212 that the prototype mainly have to run on my box.
1213 Addressing portability too early may be like optimizing too early :-/
1215 - use gras_init in the Tests instead of the crappy parse_log_opt
1216 (the latter function is removed)
1217 [Conditional execution]
1218 - New functions: gras_if_RL/gras_if_SG (basic support for this)
1219 [Code reorganisation]
1220 - Get rid of libgrasutils.a since it makes more trouble than it solves.
1221 Build examples against the RL library, since there is no way to disable
1222 its creation for now.
1224 For information, the beginning of coding on GRAS was back in june
1225 2003. I guess that every line has been rewritten at least twice since