1 SimGrid (3.1) unstable; urgency=low
5 There was a stack corruption somewhere, visible only when optimizing
6 with these versions [Vince]
9 * complete rewrote of the KCCFLN05 workstation model. It is now an
10 extension of the classical CLM03 model that gracefully handles
11 failures. This is now the standard model for MSG and GRAS. It doesn't
12 handle parallel tasks yet though. [AL]
13 * Bug fix: Weights were not correctly set in the network part.
14 WARNING: This may have resulted in incorrect results with simulations
15 where there are more than one flow on a given link. [AL]
18 * After a (long ?) discussion on simgrid-devel, we have decided that the
19 convention we had on units was stupid. That is why it has been decided
20 to move from (MBits, MFlops, seconds) fo (Bits, Flops, seconds).
21 WARNING : This means that all previous platform files will not work as
22 such with this version! A warning is issued to ask users to update
24 A conversion script can be found in the contrib module of the CVS, under
25 the name contrib/platform_generation/surfxml_update.pl [MQ]
28 * Bug fix: Processes were started in reverse order, wrt deployment file.
29 WARNING: if your code relies on this bug, please fix it. [AL]
30 * Bug fix: Add a test in MSG_task_execute to stop whenever a task is
31 being executed on two different locations. [AL]
32 * Bug fix: Failures are now better supported thanks to Derrick's tests
33 (there was many failure situations I hadn't thought of and that weren't
34 correctly handled). [AL]
35 * New function: MSG_host_is_avail indicates you whether a given m_host_t
39 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
40 Exception occuring on server-side are propagated back to client (!).
42 API CHANGE: the callback changed their prototype. Change:
43 int my_handler(gras_socket_t expeditor, void *payload_data) {
45 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
46 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
48 * New! function: gras_msg_handleall to deal with all messages arriving
49 within a given period.
50 * New! gras processes display their backtrace when they get a SIGUSR1
51 or when Ctrl-C is pressed. Use Ctrl-C Ctrl-C to exit.
52 Sweet to debug RL processes [MQ]
56 - Do not force experiment sizes to be expressed in kb, or it becomes
57 impossible to mesure the latency this way (needs one byte-long tests)
58 WARNING: this changes the amok_bw_* function semantic. [MQ]
59 - Implements the link saturation stuff. [MQ]
60 * Peer management module:
61 New! module factorizing code that we wrote over and over [MQ].
64 * New module: cunit (my jUnit implementation in ansi C) [MQ]
65 - Test units are placed directly into the library code, they get extracted
66 automatically and placed into the src/testall binary.
67 - Convert most of the XBT tests to this system.
68 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
69 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
70 live in the same C code. This required to change a little bit the API
71 of surfxml parsing but shouldn't be an issue for end-users. [AL]
72 * New module: sparse graph structure with basic algorithms (this is work
73 in progress and the API is not considered to be frozen yet). [AL]
74 * Display more information on backtraces: source line & function names are
75 now displayed just like valgrind does (rely on addr2line tool) [MQ]
76 * New function: xbt_backtrace_display(). Sweet while debuging [MQ]
77 * Reworked a little bit some #include statements to load only required
78 headers. Some user code that relied on SimGrid to include stdlib or
79 stdio may need to include it by themselves. [AL]
80 * Fixed xbt/log.h. A missing SG_BEGIN_DECL prevented compilation with
82 * Renamed xbt_host_t into xbt_peer_t since it betterly describes what I
83 meant. This breaks the API of AMOK and of xbt/config. Sorry about this,
84 but I guess that almost nobody used those parts.
88 SimGrid (3.0.1) unstable; urgency=low
91 * Unfortunately, I had missed 5 misnamed functions:
92 xbt_fifo_item_t xbt_fifo_newitem(void);
93 void xbt_fifo_freeitem(xbt_fifo_item_t);
94 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
95 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
96 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
97 They're now deprecated. Please use their new versions:
98 xbt_fifo_item_t xbt_fifo_new_item(void);
99 void xbt_fifo_free_item(xbt_fifo_item_t);
100 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
101 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
102 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
104 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
105 * Documentation: xbt_log module unmercifully reworked [MQ]
106 * Bugfix: there was a problem with the ending of contexts with
107 the pthread backend. It caused some weird deadlock or behavior
108 depending on the pthread implementation. [AL]
109 * Bugfix: get the exceptions raised in the simulator repport where
110 and why they come from when they are not catched in time [AL, MQ]
113 * Bugfix: Do repport the error when two non-connected hosts try to
114 exchange data (Thanks to Flavien for stumbling into this one) [AL]
117 * Add additionnal checkings on communications. Assert that two
118 communicating hosts are connected by a set of links... [AL]
121 * Add additionnal checkings on channel values in communication [AL]
122 * New: MSG_task_get_source to see on which host a task was generated [HC]
123 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
124 the number of tasks waiting to be received on channel and sent
126 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
127 waits for the first task coming from a given host.. [AL]
129 GRAS new functionnalities: [MQ]
130 * Enhance the parsing macro to allow the size of multidimentional objects
131 to be given thru annotations.
132 * New example (and documentation): Matrix Multiplication a la RPC
133 (as when I was young!) and fix a bunch of bugs found on the way.
135 GRAS performance improvements: [MQ]
137 * Reduce the amount of cbps creation/destruction by making it static to
138 datadesc_send/recv() and using a (newly created) cbps_reset (based on
141 * Change libdata to a set so that we can search for stuff by ID (and thus
142 reduce the insane amount of dict lookups)
145 * Actually implement gras_datadesc_copy() so that we don't have to mimick
146 RL communication on top of SG since it's so uneffective.
147 It may also be used for inter-thread communication in RL, one day.
148 * Use gras_datadesc_copy() to exchange messages on top of SG
150 - improve message exchange performance on top of SG
151 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
152 * Don't exchange on the network the size of the used part of buffer,
153 instead, specify the possible buffer size to read().
155 - reduces the amount of read/write calls (one pair per exchange)
156 - reduces the amount of exchanged data (the size)
157 - allows to retrieve all arrived data on receiver side, if we don't need
158 it right now (subsequent read will peek the buffer)
159 - allows the receiver to proceed with the begining of the stream before
160 everything is arrived
161 - make it possible to build an iov transport (using readv/writev)
163 - take care of the data with non-stable storage (like stacked data),
165 * If possible, TCP send uses vector I/O (when writev() is here)
166 - Don't use it for receive since we send data sizes and data on the
167 same stream, so we wouldn't be able to chain large amount of chunks
168 before having to flush the stuff to read the size.
169 * Rework the transport plugin mecanism to simplify it and reduce the
170 amount of pointer dereferencement when searching for the right function
173 * I guess that now, we do almost as few system calls as possible while
174 doing as few data copy as possible.
176 To improve it further, we could try to send all the sizes first and then
177 all the data (to use iov on receiving size), but it's only a partial
178 solution: when you have 2 dimensional data, the sizes of the second
179 dimension is data of the first dimension, so you need 3 streams.
181 I'm not sure the potential performance gains justify the coding burden.
183 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
185 SimGrid (3.00) stable; urgency=low
188 * New! Give the possibility to hijack the surf parser and thus bypass
189 MSG_create_environment and MSG_launch_application. Have a look at
190 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
192 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
194 SimGrid (2.96) unstable; urgency=low
199 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
200 This deprecates the xbt_error_t mecanisms.
201 It modifies (simplifies) all XBT and GRAS API.
202 MSG API keeps unchanged (exceptions raised by XBT are catched from
203 within MSG and masked with existing error handling facilities)
206 * New! Add a FATPIPE model. [AL]
207 * New! Add a parallel task model. [AL]
208 * New! Add automatically a loopback interface (in the default
209 network model) if none was precised.
212 * Bugfix: MSG_process_resume now works with the current running process.
214 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
215 * Modification of MSG_task_get_compute_duration. Once a task has been
216 processed, the value returned by this function is now equal to 0. [AL]
217 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
218 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
219 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
220 of a m_task. For now, MSG_task_cancel only works with computation
222 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
223 of the processor (in Mflop/s) regardless of the current load on the
225 * API Change: use proper naming convention for MSG_getClock and
226 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
228 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
229 This function changes the priority of a computation task. This priority
230 doesn't affect the transfer rate. A priority of 2 will make a task
231 receive two times more cpu power than the other ones. This function
232 has been added to suit the needs of Nguyen The Loc and hasn't been that
233 much tested yet. So if it fails, please report it and send me your code.
235 * API Change: removed all functions and types that were marked "deprecated"
236 since many months. Renamed MSG_global_init_args to MSG_global_init.
238 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
240 SimGrid (2.95) unstable; urgency=low
243 * Steal some nice code to GNU pth to fix context detection and usage [AL]
244 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
245 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
248 * Bugfix: Allow absolute paths for platform description files [MQ]
249 * Bugfix: do free the variables after use. Leads to drastic performance
251 * Implement max_duration (ie, timeouts) on resources [AL]
254 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
256 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
257 the needed support (timeouts) [AL]
260 * Implement measurement sockets. You can now get the bandwidth between two
261 hosts thanks to AMOK (see below). [MQ]
262 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
263 dynar over the network (yeah) [MQ]
264 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
265 * simple/static token-ring example. [Alexandre Colucci and MQ]
266 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
270 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
271 type parsing. "short" and "long" modifiers were also ignored; other
272 modifier (such as reference level) are still ignored. [MQ]
273 * Embeed the buffer size within the buffer itself on SG. [MQ]
274 That way, send() are atomic and cannot get intermixed anymore (at least
275 the ones which are less than 100k; bigger messages still have the issue)
276 * Array size pushed by the field, not by the field type (or each
277 and every long int will push stuff to the cbps) [MQ]
278 * use select() to sleep since it allows to portably sleep less than one
281 GRAS (minor cleanups)
282 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
285 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
286 * type category 'ignored' killed as it was never used and were difficult
288 * whether a type can cycle or not is now a flag, leaving room for more
289 flags (as "ignored", if we feel the need one day ;)
290 * Rename raw sockets to measurement sockets since "raw" has another
291 meaning in networking community.
294 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
295 and offers features not belonging to GRAS but that most applications
296 need. One day, it may be a set of plugins loadable at runtime.
297 * New module: bandwidth
298 bandwidth measurement between arbitrary nodes running this module. [MQ]
300 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
302 SimGrid (2.94) unstable; urgency=low
304 The first beta release of SimGrid 3 !
308 * Update the main page and the FAQ. Adding references to gforge.
311 * Add a gras_os_getpid function.
314 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
315 * Extend the logs so that they also print PID, hostname, date, ... if
317 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
318 and kill the old version which were in testsuite/
319 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
323 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
324 for users. Instead, each programming environment (GRAS, MSG,...) use it
325 when needed to provide such a feature to users.
326 Don't shortcut the mecanism or you will also shortcut the virtualization
327 you need on the simulator.
331 * Cleanups in configury with regard to compile optimization/warning flags.
332 Also add -fno-loop-optimize to any powerpc since it's the optimization
333 killing gcc (< 3.4.0).
334 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
335 needed by MSG examples complications
336 * Borrow configury beautifications from PHP
339 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
340 warning (thanks loris for stumbling into this one).
341 * Bugfix: stop loading private headers (gras_config.h) from the public
345 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
346 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
347 RL would imply a DNS resolver, which is *hard* to do in a portable way
348 (and therefore delayed).
349 * Implement a real timer mecanism and use it in timing macros. This allows
350 to avoid rounding errors and get a 0.000005 sec precision in timing
351 macros. While I was at it, various cleanups:
352 - allow to declare more than one timed section per file (fix a stupid bug)
353 - move some private declaration to the right place
354 - merge conditional execution and timing macros into emulation module
355 - document the module
356 - make sure the module cleanups its mess on gras_exit
357 * Documentation improvements:
358 - (new) how to compile applications using GRAS
359 - (new) emulation support (timing macros)
361 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
363 SimGrid (2.93) unstable; urgency=low
365 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
368 - Use Paje properly where used. Still to be sanitized properly.
369 - Portability fix: Add an implementation of the contexts using pthread
373 - Add xbt_procname(): returns the name of the current process.
374 Use it to show the current process's name in all logging.
376 - fix detection of older flex version and the reaction, since we do
377 depend on modern ones (we use lex_destroy)
378 - Better separation of SG and RL in the libs: remove all simulation code
379 from libgras. As a result, this lib is now only 200k when stripped.
380 Some of the xbt modules may also be duplicated (two sets and such) and
381 should be cleaned/killed before SG3.
382 - Insist on using xlC on AIX because of weird problems involving gcc there.
383 - Cleanup the make remote stuff. This is now done by scripts
384 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
385 mainly for our private use, but we're working on changing them to user
388 - Bugfix: flush the socket on close only if there is some *output*.
389 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
391 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
394 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
396 SimGrid (2.92) unstable; urgency=low
398 Alpha 3 on the path to SimGrid 3
402 - New! First try of benchmarking macros.
403 - New! First try so that gras_stub_generator generate deployment and
404 remote compilation helpers.
406 - Bugfix: Initialization fix in msg_test.
410 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
412 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
414 - API Change: gras_os_sleep to take the amount of seconds as a double.
415 Accepting an int was error prone since it was the only location where
416 seconds were coded as such. It leaded to damn rounding errors.
417 - Bugfix: Hard to belive that timers ever worked before this.
419 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
421 SimGrid (2.91) unstable; urgency=low
423 Alpha 2 on the path to SimGrid 3
427 - Bug fix in the lmm_solver.
429 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
430 through the function MSG_paje_output.
431 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
432 - It is possible to bound the rate of a communication in MSG with
433 MSG_task_put_bounded() (was already in the previous version but I had forgotten
434 to write it in the changelog).
435 - Bug fix to let GRAS run on top of MSG until we move it directly on top
440 - Various cleanups to the autotools stuff
441 - Begin to move Gras examples to examples/gras/
442 - Let make distcheck work again (yeah!)
444 - documentation overhauled using doxygen.
445 gtk-doc-tools is dead in SimGrid now.
446 - Automatically extract all existing logging categories, and add the list
447 to the documentation (long standing one, to say the less)
449 - Cleanup the known architecture table. Reorder the entries to group what
450 should be, and use a more consistent naming scheme.
451 (some of the test dataset are still to be regenerated)
452 - New! Allow library to register globals on each process just as userdata
454 This is implemented using a xbt_dict and not a xbt_set, so we loose the
455 lookup time (for now).
456 Use it in msg and trp.
457 This cleans a lot the internals and helps enforcing privacy of the
458 headers between the gras components.
459 - New! Add a timer mechanism, not unlike cron(8) and at(1).
460 - Bugfix: gras_os_time was delirious in RL.
461 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
463 - Reenable GRAS now that it works.
465 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
467 SimGrid (2.90) unstable; urgency=low
469 Alpha 1 on the path to SimGrid 3
471 * It is a long time since the last release of SimGrid. I'm sorry about
472 that but as I had told you, I was rewriting a lot of things. I apologize
473 to those who had been reporting bugs to me and that I had not answered.
474 If your bug is still in the new version, please tell me. Here is a
475 summary of the main changes.
477 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
478 lead by Martin Quinson. As a consequence SimGrid gains a lot in
479 portability, speed, and a lot more but you'll figure it out later.
480 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
481 I wanted to release the new MSG as soon as possible and I have
482 broken GRAS, which is the reason why, for now, only MSG is fully
483 functional. A laconic description of these projects is available
484 in the documentation.
486 * REVOLUTION 2: I have removed SG and I am now using a new simulation
487 kernel optimized for our needs (called SURF but only the developers
488 should use it). Hence, MSG is now roughly 30 times faster and I think
489 that by rewriting a little bit MSG, I could event speed it up a little
490 bit more. Beside the gain in speed, it is also much easier to encode a
491 new platform model with SURF than it was with SG. More to come...
493 * REVOLUTION 3: I have tried to change a little as possible the API of
494 MSG but a few things really had to disappear. The main differences
495 with the previous version are :
496 1) no more m_links_t and the corresponding functions. Platforms are
497 directly read from a XML description and cannot be hard-coded
498 anymore. The same format is used for application deployment
499 description. The new format is described in the documentation.
500 Have a look in tools/platform_generation. There is a tiny script
501 that converts from the old platform format to the new one. Concerning
502 the application deployment format, parsing the old one is tricky.
503 I think most of you should however be able to convert your files. If
504 it is really an issue, I can write a C code that does the conversion.
506 2) the toolbox tbx does not exist anymore. We now have a library
507 with much more data-structures but without the hash-tables (we have
508 dictionaries that are much faster).
510 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
512 *****************************************************************************
513 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
514 * don't feel like converting the oldies. (Mt) *
515 *****************************************************************************
518 Version 2.90: "the long awaited one"
519 - Finished rewriting and debugging MSG. Rewrote the documentation.
520 - disable GRAS for now since it needs to be ported to the newest SG
523 - Finish the port to windows (using mingw32 for cross-compile)
526 - Main loop and datastructures of SURF. A cpu resource object is
527 functional. Surf can thus be used to create cpu's with variable
528 performance on which you can execute some actions.
530 2004-11-15 Martin Quinson
531 - Port to ARM. Simply added the alignment and size descriptions. Should
532 work, but the ARM machines are so slow that I didn't had the opportunity
533 to 'make check' over there yet.
535 2004-11-15 Arnaud Legrand
536 - Trace manager now written. It uses a heap structure and is therefore
537 expected to be efficient. It may however be speeded up (particularly
538 when many events occur at the same date) by using red and black
539 trees. One day maybe...
540 - Max-min linear system solver written. It uses a sparse matrix
541 structure taking advantage of its expected use. Most operations are
542 O(1) and free/calloc are called as few as possible. The computation of
543 the minimum could however be improved by using a red and black tree
546 2004-11-03 Arnaud Legrand
547 - Rename every gras_* function that was in xbt/ to its xbt_
549 - Add a heap and a doubly-linked list to xbt
550 - Added a dichotomy to the dictionaries. make check works as well before
551 so I assume that the patch is correct. I do not know however if things
552 run effectively faster than before now. :)
554 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
555 SimGrid, and the version number is bumped to 2.x
557 2004-10-29 Martin Quinson
558 - Introduction of the remote errors.
559 They are the result of a RMI/RPC on the remote machine.
560 ErrCodes being scalar values, you can't get the host on which those
561 errors did happen. Extending the error mechanism as in Gnome is possible.
562 No idea yet whether it is a good idea.
564 2004-10-28 Martin Quinson
565 - Interface revolution: the Starred Structure Eradication.
566 I used to do typedef struct {} toto_t; and then handle *toto_t.
567 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
568 - ???_t is a valid type (builded with typedef)
569 - s_toto_t is a structure (access to fields with .)
570 - s_toto is a structure needing 'struct' keyword to be used
571 - e_toto_t is an enum
572 - toto_t is an 'object' (struct*)
574 typedef struct s_toto {} s_toto_t, *toto_t;
575 typedef enum {} e_toto_t;
576 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
577 s_toto_t) is private.
579 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
580 so that it can set the variable to NULL. It was so for dicts and sets,
581 it changed for dynars.
583 - Fix a bunch of memleaks in dict_remove
584 - Fix a bug in sg/server_socket opening: it failed all the time.
586 2004-10-07 Martin Quinson
587 - Speed up dynar lookup operation a bit.
589 gras_dynar_get is dead.
591 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
592 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
593 (which gives you the address of the stored data).
595 gras_dynar_get_as is an helpful macro which allows you to retrieve a
596 copy of the data using an affectation to do the job and not a memcpy.
598 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
599 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
601 It does not really speedup the dynar test because they are
602 setting elements all the time (and look them seldom). But the dict does
603 far more lookup than setting.
605 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
607 2004-10-05 Martin Quinson
608 - Allow to (en/dis)able the cycle detection at run time.
610 Whether we should check for cycle or not is now a property of each
611 datatype. When you think there may be some cycle, use datadesc_cycle_set.
612 datadesc_cycle_unset allow to remove this property when previously set.
614 Note that the cycle detection is off by default since it impacts the
615 performance. Watch the data you feed GRAS with ;)
617 This property is hereditary. Any element embedded in a structure having it
618 set have it set for the time of this data exchange.
620 You should set it both on sender and receiver side. If you don't set it on
621 sender side, it will enter an endless loop. If you forget on receiver
622 side, the cycles won't be recreated after communication.
624 - Header reorganization.
625 Kill gras_private.h, each submodule must load the headers it needs.
627 2004-10-04 Martin Quinson
628 - Interface revolution: do not try to survive to malloc failure.
630 Now, gras_malloc and friends call gras_abort() on failure.
631 As a conclusion, malloc_error is not a valid error anymore, and all
632 functions for which it was the only gras_error_t return value are
633 changed. They now return void, or there result directly.
634 This simplify the API a lot.
636 2004-09-29 Martin Quinson
637 - Re-enable raw sockets.
638 Created by gras_socket_{client,server}_ext;
639 Used with gras_raw_{send,recv}
642 It should allow to kill the last bits of gras first version soon.
644 This is not completely satisfactory yet (duplicate code with
645 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
648 - Simplify transport plugin (internal) interface by not passing any
649 argument to _server and _client, but embedding them in the socket
652 2004-09-28 Martin Quinson
653 - Finish the port to AIX.
654 autoconf was my problem (segfault within the malloc replacement
655 function. No idea why)
657 2004-09-16 Martin Quinson
658 - Fix some size_t madness on 64bit architectures.
660 2004-09-08 Martin Quinson
661 - Reduce the number of system headers loaded, overload some more system
662 calls (such as malloc to cast the result of the system one, and work
664 - Fix and reintroduce the config support
666 2004-09-07 Martin Quinson
667 - Source code reorganization to allow Arnaud to surf all over there.
668 - Allow to document the logging categories.
669 - Remove all uppercase from logging categories and useless cleanup in names.
671 2004-08-18 Martin Quinson
672 Version 0.6.2 (protocol not changed; API changed)
673 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
674 gras_error_t), and NULL when not found. Functions expecting a msgtype
675 as argument (msg_wait; msg_send) deal with NULL argument by providing a
676 hopefully usefull message.
677 - Portability to prehistoric sparcs again
679 2004-08-17 Martin Quinson
680 Version 0.6.1 (protocol not changed; ABI not changed)
681 - prealloc some buffers to speed things up
683 2004-08-11 Martin Quinson
684 Version 0.6 (protocol not changed; ABI expended)
685 - The parsing macro can deal with the references, provided that you add
686 the relevant annotations (using GRAS_ANNOTE(size,field_name))
688 2004-08-09 Martin Quinson
689 Version 0.5 (protocol not changed; ABI changed)
690 - Allow to off turn the cycle detection code in data exchange at
691 compilation time. It should be at run time, but I'm short of time (and
692 the config stuff is still broken). That way, we keep dict out of the
693 critical path, which is good because the performance is poor:
694 - search not dichotomial yet
695 - dynar give no way to access their content and memcpy everytime
696 - In composed data description (struct, ref and so on), stop foolness of
697 keeping the subtype's ID, but store the type itself. This keeps sets out
698 of the critical path, which is good since they rely on dynar and
699 dictionnaries. The only loose of that is that we cannot detect the
700 redeclaration of a structure/union with another content (but I'm not sure
701 the code detected well this error before anyway). We still can detect
702 the redefinition discrepancy for the other types.
703 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
704 it breaks the code because of type-punning used all over the place).
705 This breaks on all non-gcc architectures (for now).
707 All those changes (plus the buffer of last time) allow me to gain 2 order
708 of magnitude on cruel tests consisting of 800000 array of integers on two
709 level of a hierarchical structure (200 secondes -> 4 secondes)
712 - the selector of reference must now return the type it points to, not
715 2004-08-06 Martin Quinson
716 Version 0.4 (protocol changed; ABI not changed)
717 - Allow to pass --gras-log argument to processes in simulation mode. Really.
718 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
719 - Implement a buffer transport, and use it by default (it relies on tcp in
720 real life and on sg in simulation).
721 That's a bit hackish since I had a new field to the structure to store
722 its data without interfering with the subtype ones. Inheritance
723 is tricky in C. And that's a kind of reverse inheritance with one class
724 derivating two classes. Or maybe a game with java interfaces. Anyway,
725 that's damn hard in C (at least).
726 Moreover, I got tired while trying to ensure plugin separation and
727 genericity in SG mode. MSG wants me to do weird things, so let's go for
728 cruel hacks (temporarily of course ;).
729 See comment in transport_private.h:71
730 - do not include all the _interface headers in private but in the files
731 which really need them (to cut the compilation time when they are
734 2004-07-26 Martin Quinson
735 Version 0.3 (protocol not changed; ABI changed)
736 - Major overhault of the datadesc interface to simplify it:
737 - shorted the function names:
738 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
739 - add a trivial way to push/pop integers into the cbps without malloc.
740 This allows to make really generic sub_type description, which simply
741 pop their size of the stack.
742 - add a function gras_datadesc_ref_pop_arr() which does what users want
743 most of the time: Declare a dynamic array (which pops its size of the
744 stack) and declare a reference to it. Poor name, but anyway.
745 - kill the post-send callback, add a post-receive one
747 2004-07-23 Martin Quinson
748 Version 0.2 (protocol changed; ABI changed)
749 - add some testing for cpbs in the test cases, and fix some more bugs.
750 This invalidate again the little64 data file, since I cannot regenerate
752 - remove an awfull optimization in the logging stuff, allowing me to:
753 - understand it again
754 - learn gcc how to check that the argument match the provided format
755 - fix all errors revealed by gcc after that
756 - internal keys of dict are not \0 terminated. Deal with it properly in
757 loggings instead of segfaulting when the user want to see the logs :-/
759 2004-07-22 Martin Quinson
760 - Fix some stupid bug preventing cbps (callback postit) from working
762 2004-07-21 Martin Quinson
763 - Some documentation cleanups
764 - remove the useless last argument of msgtype_declare
765 - rename the Virtu functions to fit into the 'os' namespace
766 - move headers src/include -> src/include/gras/ and stop fooling with
767 gras -> . symbolic link
768 - make distcheck is now successful
770 2004-07-19 Martin Quinson
772 - Build shared library also
773 - Install html doc to the right location
774 - stop removing maintainer files in make clean
775 - build tests only on make check
777 2004-07-13 Martin Quinson
779 - No major issue in previous version => change versionning schema
780 - Re-enable little64 convertion test now that Abdou kindly regenerated the
781 corresponding dataset.
783 2004-07-11 Martin Quinson
785 - Get it working with any kind of structure (we can compute the padding
786 bytes remotely for all the architectures I have access to)
787 - Implement the structure parsing macro (still not quite robust/complete)
788 - Improvement to the remote testing toysuite
790 2004-07-10 Martin Quinson
792 - get ride of a bunch of deprecated macros
793 - actually run the test for two-compliment, not only compile it :-/
794 - test whether the structures get packed (and bail out if yes. Damn.
795 Alignment is a serious matter)
796 - test whether the structures get compacted (but respecting the alignment
797 constraints of each types)
798 - test whether the array fields of structures can straddle alignment boundaries
800 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
801 (took me 2 hours to find that bug, looking at the wrong place)
802 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
803 before sending/receiving and are used to compute the offsets of fields
804 - Given that padding size depend even on compiler options, keep track of
805 alignment and aligned_size only for the current architecture. Not a big
806 deal since we send structure fields one after the other (seems
808 - Add the datastructure used for IEEE paper by the PBIO guys to the test
809 program, let it work on linux/gcc/little32. portability todo.
811 2004-07-08 Martin Quinson
812 - import and improve remote compilation support from FAST
813 - make sure make check works on half a dozen of machines out there
815 2004-07-07 Martin Quinson
816 Let's say it's version 0.0.3 ;)
817 - Implement conversions (yuhu!)
818 - Let it work on solaris (beside conversion, of course)
819 - Stupid me, using rand() to generate the conversion datatests in not wise.
821 2004-07-06 Martin Quinson
822 - Let make dist work, since I'm gonna need it to compile on remote hosts
823 - Let Tests/datadesc_usage write the architecture on which the file was
824 generated as first byte.
825 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
826 architecture descriptions.
827 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
828 on those architectures.
829 - Optimization: send/recv array of scalar in one shoot
831 2004-07-05 Martin Quinson
832 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
834 - Plug a whole bunch of memleaks
835 - each process now have to call gras_{init,exit}. One day, their log
836 settings will be separated
837 - Continue the code factorisation between SG, RL and common in Transport.
839 2004-07-04 Martin Quinson
841 - Redistribution between SG and RL.
842 We wanna have to accept in SG, so move accepted related parts of RL in
843 the common part. (more precisely, the dynar of all known sockets is no
844 more a static in transport.c, but part of the process_data)
846 [gras_stub_generator]
847 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
849 2004-07-03 Martin Quinson
850 - Create a new log channel tbx containing dict, set, log, dynar (to shut
851 them all up in one shot)
853 - Fix the ugly case of reference to dynamic array.
854 - New (semi-public) function gras_datadesc_size to allow the messaging
855 layer to malloc the needed space for the buffer.
857 - gras_socket_close now expect the socket to close (and not its address to
858 put NULL in it after it). This is because the socket passed to handlers
859 is one of their argument (=> not writable).
861 - propagate the interface cleanup from last week in datadesc, ie remove a
862 superfluous level of indirection. User pass adress of variable
863 containing data (both when sending and receiving), and not of a variable
864 being a pointer to the data. Let's say that I like it better ;)
865 The price for that is constructs like "int msg=*(int*)payload" in
866 handlers, but it's a fine price, IMHO.
868 - Let it work in RL (yuhu)
870 2004-06-21 Martin Quinson
872 - porting SG plugin and SG select to new standards (works almost).
873 - plug memleaks and fix bugs around.
876 - cleanup the prototype of data recv and force users to specify when they
877 want to handle references to objects. Test case working even for cycles.
878 - plug memleaks. Valgrind is perfectly ok with this.
880 2004-06-12 Martin Quinson
882 - cleanup the separation between plugin and main code in plugin creation
884 2004-06-11 Martin Quinson
886 - Reput hook for raw sockets, needed for BW experiments
887 - kill a few lines of dead code
888 [Data description] Interface cleanup
889 - gras_datadesc_by_name returns the searched type or NULL.
890 That way, no variable is needed to use a type desc once, which makes
892 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
893 two parameters were strings (field name, type name), leading to
895 [Dicos] Interface cleanup
896 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
897 This is consistant with the dynar API.
899 2004-04-21 Martin Quinson
901 - Porting to new standards.
904 There is no bag anymore, no need to take extra provision to mask the
905 pointers behind "ID".
906 Better splitup of functions between files create/exchange/convert.
907 This is still a bit artificial since convert and receive are so
908 interleaved, but anyway.
910 - add a queued message list to procdata (the ones not matching criteria
912 - factorize some more code between SG and RL wrt procdata
914 - use gras_exit in example to track memleaks
915 - get rid of gs_example now that GS is properly integrated into gras
916 - update run_test to integrate the lastest tests (datadesc)
918 - rename WARNINGn macros to WARNn since it prooved error-prone
920 2004-04-19 Martin Quinson
922 - register init/exit functions within gras module mechanism
923 - send/receive function.
924 Convertion is not implemented, but short-cutted if not needed.
925 struct/array elements are sent one by one (instead of block-wise), but
926 nobody really cares (yet). Get a prototype before optimizing.
927 - tests (using a file socket) for DD send/receive on:
928 - base types: int, float
929 - array: fixed size, string (ie ref to dynamic string)
930 - structure: homogeneous, heterogeneous
931 - chained list, graph with cycle
932 Believe it or not, valgrind is not too unhappy with the results. The
933 cycle happily segfaults, but the others are ok. And I'm sick of pointers
937 - Bugfix when using a filename explicitely (instead of '-')
939 2004-04-09 Martin Quinson
941 - factorize more code between RL and SG in socket creation
942 - Complete the implementation and tests of:
944 o file (only in RL, and mainly for debugging)
946 I lost 3 days to design a portable address resolver, and then decided
947 that the prototype mainly have to run on my box.
948 Addressing portability too early may be like optimizing too early :-/
950 - use gras_init in the Tests instead of the crappy parse_log_opt
951 (the latter function is removed)
952 [Conditional execution]
953 - New functions: gras_if_RL/gras_if_SG (basic support for this)
954 [Code reorganisation]
955 - Get rid of libgrasutils.a since it makes more trouble than it solves.
956 Build examples against the RL library, since there is no way to disable
957 its creation for now.