1 SimGrid (3.0.2) unstable; urgency=low
4 * New module: cunit (my jUnit implementation in ansi C) [MQ]
5 - Test units are placed directly into the library code, they get extracted
6 automatically and placed into the src/testall binary.
7 - Convert most of the XBT tests to this system.
8 * New functions: xbt_dynar_getfirst_as() and xbt_dynar_getlast_as() [MQ]
9 * XML parsing: rewrote parts of flexml to enable multiple xml parsers to
10 live in the same C code. This required to change a little bit the API
11 of surfxml parsing but shouldn't be an issue for end-users. [AL]
12 * New module: sparse graph structure with basic algorithms. [AL]
15 * complete rewrote of the KCCFLN05 workstation model. It is now an
16 extension of the classical CLM03 model that gracefully handles
17 failures. This is now the standard model for MSG and GRAS. [AL]
20 * Bug fix: Processes were started in reverse order, wrt deployment file.
21 WARNING: if your code relies on this bug, please fix it. [AL]
25 - Do not force experiment sizes to be expressed in kb, or it becomes
26 impossible to mesure the latency this way (needs one byte-long tests)
27 WARNING: this changes the amok_bw_* function semantic.
31 SimGrid (3.0.1) unstable; urgency=low
34 * Unfortunately, I had missed 5 misnamed functions:
35 xbt_fifo_item_t xbt_fifo_newitem(void);
36 void xbt_fifo_freeitem(xbt_fifo_item_t);
37 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
38 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
39 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
40 They're now deprecated. Please use their new versions:
41 xbt_fifo_item_t xbt_fifo_new_item(void);
42 void xbt_fifo_free_item(xbt_fifo_item_t);
43 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
44 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
45 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
47 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
48 * Documentation: xbt_log module unmercifully reworked [MQ]
49 * Bugfix: there was a problem with the ending of contexts with
50 the pthread backend. It caused some weird deadlock or behavior
51 depending on the pthread implementation. [AL]
52 * Bugfix: get the exceptions raised in the simulator repport where
53 and why they come from when they are not catched in time [AL, MQ]
56 * Bugfix: Do repport the error when two non-connected hosts try to
57 exchange data (Thanks to Flavien for stumbling into this one) [AL]
60 * Add additionnal checkings on communications. Assert that two
61 communicating hosts are connected by a set of links... [AL]
64 * Add additionnal checkings on channel values in communication [AL]
65 * New: MSG_task_get_source to see on which host a task was generated [HC]
66 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
67 the number of tasks waiting to be received on channel and sent
69 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
70 waits for the first task coming from a given host.. [AL]
72 GRAS new functionnalities: [MQ]
73 * Enhance the parsing macro to allow the size of multidimentional objects
74 to be given thru annotations.
75 * New example (and documentation): Matrix Multiplication a la RPC
76 (as when I was young!) and fix a bunch of bugs found on the way.
78 GRAS performance improvements: [MQ]
80 * Reduce the amount of cbps creation/destruction by making it static to
81 datadesc_send/recv() and using a (newly created) cbps_reset (based on
84 * Change libdata to a set so that we can search for stuff by ID (and thus
85 reduce the insane amount of dict lookups)
88 * Actually implement gras_datadesc_copy() so that we don't have to mimick
89 RL communication on top of SG since it's so uneffective.
90 It may also be used for inter-thread communication in RL, one day.
91 * Use gras_datadesc_copy() to exchange messages on top of SG
93 - improve message exchange performance on top of SG
94 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
95 * Don't exchange on the network the size of the used part of buffer,
96 instead, specify the possible buffer size to read().
98 - reduces the amount of read/write calls (one pair per exchange)
99 - reduces the amount of exchanged data (the size)
100 - allows to retrieve all arrived data on receiver side, if we don't need
101 it right now (subsequent read will peek the buffer)
102 - allows the receiver to proceed with the begining of the stream before
103 everything is arrived
104 - make it possible to build an iov transport (using readv/writev)
106 - take care of the data with non-stable storage (like stacked data),
108 * If possible, TCP send uses vector I/O (when writev() is here)
109 - Don't use it for receive since we send data sizes and data on the
110 same stream, so we wouldn't be able to chain large amount of chunks
111 before having to flush the stuff to read the size.
112 * Rework the transport plugin mecanism to simplify it and reduce the
113 amount of pointer dereferencement when searching for the right function
116 * I guess that now, we do almost as few system calls as possible while
117 doing as few data copy as possible.
119 To improve it further, we could try to send all the sizes first and then
120 all the data (to use iov on receiving size), but it's only a partial
121 solution: when you have 2 dimensional data, the sizes of the second
122 dimension is data of the first dimension, so you need 3 streams.
124 I'm not sure the potential performance gains justify the coding burden.
126 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
128 SimGrid (3.00) stable; urgency=low
131 * New! Give the possibility to hijack the surf parser and thus bypass
132 MSG_create_environment and MSG_launch_application. Have a look at
133 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
135 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
137 SimGrid (2.96) unstable; urgency=low
142 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
143 This deprecates the xbt_error_t mecanisms.
144 It modifies (simplifies) all XBT and GRAS API.
145 MSG API keeps unchanged (exceptions raised by XBT are catched from
146 within MSG and masked with existing error handling facilities)
149 * New! Add a FATPIPE model. [AL]
150 * New! Add a parallel task model. [AL]
151 * New! Add automatically a loopback interface (in the default
152 network model) if none was precised.
155 * Bugfix: MSG_process_resume now works with the current running process.
157 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
158 * Modification of MSG_task_get_compute_duration. Once a task has been
159 processed, the value returned by this function is now equal to 0. [AL]
160 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
161 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
162 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
163 of a m_task. For now, MSG_task_cancel only works with computation
165 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
166 of the processor (in Mflop/s) regardless of the current load on the
168 * API Change: use proper naming convention for MSG_getClock and
169 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
171 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
172 This function changes the priority of a computation task. This priority
173 doesn't affect the transfer rate. A priority of 2 will make a task
174 receive two times more cpu power than the other ones. This function
175 has been added to suit the needs of Nguyen The Loc and hasn't been that
176 much tested yet. So if it fails, please report it and send me your code.
178 * API Change: removed all functions and types that were marked "deprecated"
179 since many months. Renamed MSG_global_init_args to MSG_global_init.
181 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
183 SimGrid (2.95) unstable; urgency=low
186 * Steal some nice code to GNU pth to fix context detection and usage [AL]
187 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
188 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
191 * Bugfix: Allow absolute paths for platform description files [MQ]
192 * Bugfix: do free the variables after use. Leads to drastic performance
194 * Implement max_duration (ie, timeouts) on resources [AL]
197 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
199 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
200 the needed support (timeouts) [AL]
203 * Implement measurement sockets. You can now get the bandwidth between two
204 hosts thanks to AMOK (see below). [MQ]
205 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
206 dynar over the network (yeah) [MQ]
207 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
208 * simple/static token-ring example. [Alexandre Colucci and MQ]
209 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
213 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
214 type parsing. "short" and "long" modifiers were also ignored; other
215 modifier (such as reference level) are still ignored. [MQ]
216 * Embeed the buffer size within the buffer itself on SG. [MQ]
217 That way, send() are atomic and cannot get intermixed anymore (at least
218 the ones which are less than 100k; bigger messages still have the issue)
219 * Array size pushed by the field, not by the field type (or each
220 and every long int will push stuff to the cbps) [MQ]
221 * use select() to sleep since it allows to portably sleep less than one
224 GRAS (minor cleanups)
225 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
228 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
229 * type category 'ignored' killed as it was never used and were difficult
231 * whether a type can cycle or not is now a flag, leaving room for more
232 flags (as "ignored", if we feel the need one day ;)
233 * Rename raw sockets to measurement sockets since "raw" has another
234 meaning in networking community.
237 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
238 and offers features not belonging to GRAS but that most applications
239 need. One day, it may be a set of plugins loadable at runtime.
240 * New module: bandwidth
241 bandwidth measurement between arbitrary nodes running this module. [MQ]
243 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
245 SimGrid (2.94) unstable; urgency=low
247 The first beta release of SimGrid 3 !
251 * Update the main page and the FAQ. Adding references to gforge.
254 * Add a gras_os_getpid function.
257 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
258 * Extend the logs so that they also print PID, hostname, date, ... if
260 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
261 and kill the old version which were in testsuite/
262 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
266 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
267 for users. Instead, each programming environment (GRAS, MSG,...) use it
268 when needed to provide such a feature to users.
269 Don't shortcut the mecanism or you will also shortcut the virtualization
270 you need on the simulator.
274 * Cleanups in configury with regard to compile optimization/warning flags.
275 Also add -fno-loop-optimize to any powerpc since it's the optimization
276 killing gcc (< 3.4.0).
277 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
278 needed by MSG examples complications
279 * Borrow configury beautifications from PHP
282 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
283 warning (thanks loris for stumbling into this one).
284 * Bugfix: stop loading private headers (gras_config.h) from the public
288 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
289 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
290 RL would imply a DNS resolver, which is *hard* to do in a portable way
291 (and therefore delayed).
292 * Implement a real timer mecanism and use it in timing macros. This allows
293 to avoid rounding errors and get a 0.000005 sec precision in timing
294 macros. While I was at it, various cleanups:
295 - allow to declare more than one timed section per file (fix a stupid bug)
296 - move some private declaration to the right place
297 - merge conditional execution and timing macros into emulation module
298 - document the module
299 - make sure the module cleanups its mess on gras_exit
300 * Documentation improvements:
301 - (new) how to compile applications using GRAS
302 - (new) emulation support (timing macros)
304 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
306 SimGrid (2.93) unstable; urgency=low
308 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
311 - Use Paje properly where used. Still to be sanitized properly.
312 - Portability fix: Add an implementation of the contexts using pthread
316 - Add xbt_procname(): returns the name of the current process.
317 Use it to show the current process's name in all logging.
319 - fix detection of older flex version and the reaction, since we do
320 depend on modern ones (we use lex_destroy)
321 - Better separation of SG and RL in the libs: remove all simulation code
322 from libgras. As a result, this lib is now only 200k when stripped.
323 Some of the xbt modules may also be duplicated (two sets and such) and
324 should be cleaned/killed before SG3.
325 - Insist on using xlC on AIX because of weird problems involving gcc there.
326 - Cleanup the make remote stuff. This is now done by scripts
327 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
328 mainly for our private use, but we're working on changing them to user
331 - Bugfix: flush the socket on close only if there is some *output*.
332 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
334 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
337 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
339 SimGrid (2.92) unstable; urgency=low
341 Alpha 3 on the path to SimGrid 3
345 - New! First try of benchmarking macros.
346 - New! First try so that gras_stub_generator generate deployment and
347 remote compilation helpers.
349 - Bugfix: Initialization fix in msg_test.
353 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
355 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
357 - API Change: gras_os_sleep to take the amount of seconds as a double.
358 Accepting an int was error prone since it was the only location where
359 seconds were coded as such. It leaded to damn rounding errors.
360 - Bugfix: Hard to belive that timers ever worked before this.
362 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
364 SimGrid (2.91) unstable; urgency=low
366 Alpha 2 on the path to SimGrid 3
370 - Bug fix in the lmm_solver.
372 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
373 through the function MSG_paje_output.
374 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
375 - It is possible to bound the rate of a communication in MSG with
376 MSG_task_put_bounded() (was already in the previous version but I had forgotten
377 to write it in the changelog).
378 - Bug fix to let GRAS run on top of MSG until we move it directly on top
383 - Various cleanups to the autotools stuff
384 - Begin to move Gras examples to examples/gras/
385 - Let make distcheck work again (yeah!)
387 - documentation overhauled using doxygen.
388 gtk-doc-tools is dead in SimGrid now.
389 - Automatically extract all existing logging categories, and add the list
390 to the documentation (long standing one, to say the less)
392 - Cleanup the known architecture table. Reorder the entries to group what
393 should be, and use a more consistent naming scheme.
394 (some of the test dataset are still to be regenerated)
395 - New! Allow library to register globals on each process just as userdata
397 This is implemented using a xbt_dict and not a xbt_set, so we loose the
398 lookup time (for now).
399 Use it in msg and trp.
400 This cleans a lot the internals and helps enforcing privacy of the
401 headers between the gras components.
402 - New! Add a timer mechanism, not unlike cron(8) and at(1).
403 - Bugfix: gras_os_time was delirious in RL.
404 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
406 - Reenable GRAS now that it works.
408 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
410 SimGrid (2.90) unstable; urgency=low
412 Alpha 1 on the path to SimGrid 3
414 * It is a long time since the last release of SimGrid. I'm sorry about
415 that but as I had told you, I was rewriting a lot of things. I apologize
416 to those who had been reporting bugs to me and that I had not answered.
417 If your bug is still in the new version, please tell me. Here is a
418 summary of the main changes.
420 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
421 lead by Martin Quinson. As a consequence SimGrid gains a lot in
422 portability, speed, and a lot more but you'll figure it out later.
423 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
424 I wanted to release the new MSG as soon as possible and I have
425 broken GRAS, which is the reason why, for now, only MSG is fully
426 functional. A laconic description of these projects is available
427 in the documentation.
429 * REVOLUTION 2: I have removed SG and I am now using a new simulation
430 kernel optimized for our needs (called SURF but only the developers
431 should use it). Hence, MSG is now roughly 30 times faster and I think
432 that by rewriting a little bit MSG, I could event speed it up a little
433 bit more. Beside the gain in speed, it is also much easier to encode a
434 new platform model with SURF than it was with SG. More to come...
436 * REVOLUTION 3: I have tried to change a little as possible the API of
437 MSG but a few things really had to disappear. The main differences
438 with the previous version are :
439 1) no more m_links_t and the corresponding functions. Platforms are
440 directly read from a XML description and cannot be hard-coded
441 anymore. The same format is used for application deployment
442 description. The new format is described in the documentation.
443 Have a look in tools/platform_generation. There is a tiny script
444 that converts from the old platform format to the new one. Concerning
445 the application deployment format, parsing the old one is tricky.
446 I think most of you should however be able to convert your files. If
447 it is really an issue, I can write a C code that does the conversion.
449 2) the toolbox tbx does not exist anymore. We now have a library
450 with much more data-structures but without the hash-tables (we have
451 dictionaries that are much faster).
453 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
455 *****************************************************************************
456 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
457 * don't feel like converting the oldies. (Mt) *
458 *****************************************************************************
461 Version 2.90: "the long awaited one"
462 - Finished rewriting and debugging MSG. Rewrote the documentation.
463 - disable GRAS for now since it needs to be ported to the newest SG
466 - Finish the port to windows (using mingw32 for cross-compile)
469 - Main loop and datastructures of SURF. A cpu resource object is
470 functional. Surf can thus be used to create cpu's with variable
471 performance on which you can execute some actions.
473 2004-11-15 Martin Quinson
474 - Port to ARM. Simply added the alignment and size descriptions. Should
475 work, but the ARM machines are so slow that I didn't had the opportunity
476 to 'make check' over there yet.
478 2004-11-15 Arnaud Legrand
479 - Trace manager now written. It uses a heap structure and is therefore
480 expected to be efficient. It may however be speeded up (particularly
481 when many events occur at the same date) by using red and black
482 trees. One day maybe...
483 - Max-min linear system solver written. It uses a sparse matrix
484 structure taking advantage of its expected use. Most operations are
485 O(1) and free/calloc are called as few as possible. The computation of
486 the minimum could however be improved by using a red and black tree
489 2004-11-03 Arnaud Legrand
490 - Rename every gras_* function that was in xbt/ to its xbt_
492 - Add a heap and a doubly-linked list to xbt
493 - Added a dichotomy to the dictionaries. make check works as well before
494 so I assume that the patch is correct. I do not know however if things
495 run effectively faster than before now. :)
497 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
498 SimGrid, and the version number is bumped to 2.x
500 2004-10-29 Martin Quinson
501 - Introduction of the remote errors.
502 They are the result of a RMI/RPC on the remote machine.
503 ErrCodes being scalar values, you can't get the host on which those
504 errors did happen. Extending the error mechanism as in Gnome is possible.
505 No idea yet whether it is a good idea.
507 2004-10-28 Martin Quinson
508 - Interface revolution: the Starred Structure Eradication.
509 I used to do typedef struct {} toto_t; and then handle *toto_t.
510 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
511 - ???_t is a valid type (builded with typedef)
512 - s_toto_t is a structure (access to fields with .)
513 - s_toto is a structure needing 'struct' keyword to be used
514 - e_toto_t is an enum
515 - toto_t is an 'object' (struct*)
517 typedef struct s_toto {} s_toto_t, *toto_t;
518 typedef enum {} e_toto_t;
519 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
520 s_toto_t) is private.
522 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
523 so that it can set the variable to NULL. It was so for dicts and sets,
524 it changed for dynars.
526 - Fix a bunch of memleaks in dict_remove
527 - Fix a bug in sg/server_socket opening: it failed all the time.
529 2004-10-07 Martin Quinson
530 - Speed up dynar lookup operation a bit.
532 gras_dynar_get is dead.
534 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
535 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
536 (which gives you the address of the stored data).
538 gras_dynar_get_as is an helpful macro which allows you to retrieve a
539 copy of the data using an affectation to do the job and not a memcpy.
541 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
542 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
544 It does not really speedup the dynar test because they are
545 setting elements all the time (and look them seldom). But the dict does
546 far more lookup than setting.
548 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
550 2004-10-05 Martin Quinson
551 - Allow to (en/dis)able the cycle detection at run time.
553 Whether we should check for cycle or not is now a property of each
554 datatype. When you think there may be some cycle, use datadesc_cycle_set.
555 datadesc_cycle_unset allow to remove this property when previously set.
557 Note that the cycle detection is off by default since it impacts the
558 performance. Watch the data you feed GRAS with ;)
560 This property is hereditary. Any element embedded in a structure having it
561 set have it set for the time of this data exchange.
563 You should set it both on sender and receiver side. If you don't set it on
564 sender side, it will enter an endless loop. If you forget on receiver
565 side, the cycles won't be recreated after communication.
567 - Header reorganization.
568 Kill gras_private.h, each submodule must load the headers it needs.
570 2004-10-04 Martin Quinson
571 - Interface revolution: do not try to survive to malloc failure.
573 Now, gras_malloc and friends call gras_abort() on failure.
574 As a conclusion, malloc_error is not a valid error anymore, and all
575 functions for which it was the only gras_error_t return value are
576 changed. They now return void, or there result directly.
577 This simplify the API a lot.
579 2004-09-29 Martin Quinson
580 - Re-enable raw sockets.
581 Created by gras_socket_{client,server}_ext;
582 Used with gras_raw_{send,recv}
585 It should allow to kill the last bits of gras first version soon.
587 This is not completely satisfactory yet (duplicate code with
588 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
591 - Simplify transport plugin (internal) interface by not passing any
592 argument to _server and _client, but embedding them in the socket
595 2004-09-28 Martin Quinson
596 - Finish the port to AIX.
597 autoconf was my problem (segfault within the malloc replacement
598 function. No idea why)
600 2004-09-16 Martin Quinson
601 - Fix some size_t madness on 64bit architectures.
603 2004-09-08 Martin Quinson
604 - Reduce the number of system headers loaded, overload some more system
605 calls (such as malloc to cast the result of the system one, and work
607 - Fix and reintroduce the config support
609 2004-09-07 Martin Quinson
610 - Source code reorganization to allow Arnaud to surf all over there.
611 - Allow to document the logging categories.
612 - Remove all uppercase from logging categories and useless cleanup in names.
614 2004-08-18 Martin Quinson
615 Version 0.6.2 (protocol not changed; API changed)
616 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
617 gras_error_t), and NULL when not found. Functions expecting a msgtype
618 as argument (msg_wait; msg_send) deal with NULL argument by providing a
619 hopefully usefull message.
620 - Portability to prehistoric sparcs again
622 2004-08-17 Martin Quinson
623 Version 0.6.1 (protocol not changed; ABI not changed)
624 - prealloc some buffers to speed things up
626 2004-08-11 Martin Quinson
627 Version 0.6 (protocol not changed; ABI expended)
628 - The parsing macro can deal with the references, provided that you add
629 the relevant annotations (using GRAS_ANNOTE(size,field_name))
631 2004-08-09 Martin Quinson
632 Version 0.5 (protocol not changed; ABI changed)
633 - Allow to off turn the cycle detection code in data exchange at
634 compilation time. It should be at run time, but I'm short of time (and
635 the config stuff is still broken). That way, we keep dict out of the
636 critical path, which is good because the performance is poor:
637 - search not dichotomial yet
638 - dynar give no way to access their content and memcpy everytime
639 - In composed data description (struct, ref and so on), stop foolness of
640 keeping the subtype's ID, but store the type itself. This keeps sets out
641 of the critical path, which is good since they rely on dynar and
642 dictionnaries. The only loose of that is that we cannot detect the
643 redeclaration of a structure/union with another content (but I'm not sure
644 the code detected well this error before anyway). We still can detect
645 the redefinition discrepancy for the other types.
646 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
647 it breaks the code because of type-punning used all over the place).
648 This breaks on all non-gcc architectures (for now).
650 All those changes (plus the buffer of last time) allow me to gain 2 order
651 of magnitude on cruel tests consisting of 800000 array of integers on two
652 level of a hierarchical structure (200 secondes -> 4 secondes)
655 - the selector of reference must now return the type it points to, not
658 2004-08-06 Martin Quinson
659 Version 0.4 (protocol changed; ABI not changed)
660 - Allow to pass --gras-log argument to processes in simulation mode. Really.
661 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
662 - Implement a buffer transport, and use it by default (it relies on tcp in
663 real life and on sg in simulation).
664 That's a bit hackish since I had a new field to the structure to store
665 its data without interfering with the subtype ones. Inheritance
666 is tricky in C. And that's a kind of reverse inheritance with one class
667 derivating two classes. Or maybe a game with java interfaces. Anyway,
668 that's damn hard in C (at least).
669 Moreover, I got tired while trying to ensure plugin separation and
670 genericity in SG mode. MSG wants me to do weird things, so let's go for
671 cruel hacks (temporarily of course ;).
672 See comment in transport_private.h:71
673 - do not include all the _interface headers in private but in the files
674 which really need them (to cut the compilation time when they are
677 2004-07-26 Martin Quinson
678 Version 0.3 (protocol not changed; ABI changed)
679 - Major overhault of the datadesc interface to simplify it:
680 - shorted the function names:
681 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
682 - add a trivial way to push/pop integers into the cbps without malloc.
683 This allows to make really generic sub_type description, which simply
684 pop their size of the stack.
685 - add a function gras_datadesc_ref_pop_arr() which does what users want
686 most of the time: Declare a dynamic array (which pops its size of the
687 stack) and declare a reference to it. Poor name, but anyway.
688 - kill the post-send callback, add a post-receive one
690 2004-07-23 Martin Quinson
691 Version 0.2 (protocol changed; ABI changed)
692 - add some testing for cpbs in the test cases, and fix some more bugs.
693 This invalidate again the little64 data file, since I cannot regenerate
695 - remove an awfull optimization in the logging stuff, allowing me to:
696 - understand it again
697 - learn gcc how to check that the argument match the provided format
698 - fix all errors revealed by gcc after that
699 - internal keys of dict are not \0 terminated. Deal with it properly in
700 loggings instead of segfaulting when the user want to see the logs :-/
702 2004-07-22 Martin Quinson
703 - Fix some stupid bug preventing cbps (callback postit) from working
705 2004-07-21 Martin Quinson
706 - Some documentation cleanups
707 - remove the useless last argument of msgtype_declare
708 - rename the Virtu functions to fit into the 'os' namespace
709 - move headers src/include -> src/include/gras/ and stop fooling with
710 gras -> . symbolic link
711 - make distcheck is now successful
713 2004-07-19 Martin Quinson
715 - Build shared library also
716 - Install html doc to the right location
717 - stop removing maintainer files in make clean
718 - build tests only on make check
720 2004-07-13 Martin Quinson
722 - No major issue in previous version => change versionning schema
723 - Re-enable little64 convertion test now that Abdou kindly regenerated the
724 corresponding dataset.
726 2004-07-11 Martin Quinson
728 - Get it working with any kind of structure (we can compute the padding
729 bytes remotely for all the architectures I have access to)
730 - Implement the structure parsing macro (still not quite robust/complete)
731 - Improvement to the remote testing toysuite
733 2004-07-10 Martin Quinson
735 - get ride of a bunch of deprecated macros
736 - actually run the test for two-compliment, not only compile it :-/
737 - test whether the structures get packed (and bail out if yes. Damn.
738 Alignment is a serious matter)
739 - test whether the structures get compacted (but respecting the alignment
740 constraints of each types)
741 - test whether the array fields of structures can straddle alignment boundaries
743 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
744 (took me 2 hours to find that bug, looking at the wrong place)
745 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
746 before sending/receiving and are used to compute the offsets of fields
747 - Given that padding size depend even on compiler options, keep track of
748 alignment and aligned_size only for the current architecture. Not a big
749 deal since we send structure fields one after the other (seems
751 - Add the datastructure used for IEEE paper by the PBIO guys to the test
752 program, let it work on linux/gcc/little32. portability todo.
754 2004-07-08 Martin Quinson
755 - import and improve remote compilation support from FAST
756 - make sure make check works on half a dozen of machines out there
758 2004-07-07 Martin Quinson
759 Let's say it's version 0.0.3 ;)
760 - Implement conversions (yuhu!)
761 - Let it work on solaris (beside conversion, of course)
762 - Stupid me, using rand() to generate the conversion datatests in not wise.
764 2004-07-06 Martin Quinson
765 - Let make dist work, since I'm gonna need it to compile on remote hosts
766 - Let Tests/datadesc_usage write the architecture on which the file was
767 generated as first byte.
768 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
769 architecture descriptions.
770 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
771 on those architectures.
772 - Optimization: send/recv array of scalar in one shoot
774 2004-07-05 Martin Quinson
775 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
777 - Plug a whole bunch of memleaks
778 - each process now have to call gras_{init,exit}. One day, their log
779 settings will be separated
780 - Continue the code factorisation between SG, RL and common in Transport.
782 2004-07-04 Martin Quinson
784 - Redistribution between SG and RL.
785 We wanna have to accept in SG, so move accepted related parts of RL in
786 the common part. (more precisely, the dynar of all known sockets is no
787 more a static in transport.c, but part of the process_data)
789 [gras_stub_generator]
790 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
792 2004-07-03 Martin Quinson
793 - Create a new log channel tbx containing dict, set, log, dynar (to shut
794 them all up in one shot)
796 - Fix the ugly case of reference to dynamic array.
797 - New (semi-public) function gras_datadesc_size to allow the messaging
798 layer to malloc the needed space for the buffer.
800 - gras_socket_close now expect the socket to close (and not its address to
801 put NULL in it after it). This is because the socket passed to handlers
802 is one of their argument (=> not writable).
804 - propagate the interface cleanup from last week in datadesc, ie remove a
805 superfluous level of indirection. User pass adress of variable
806 containing data (both when sending and receiving), and not of a variable
807 being a pointer to the data. Let's say that I like it better ;)
808 The price for that is constructs like "int msg=*(int*)payload" in
809 handlers, but it's a fine price, IMHO.
811 - Let it work in RL (yuhu)
813 2004-06-21 Martin Quinson
815 - porting SG plugin and SG select to new standards (works almost).
816 - plug memleaks and fix bugs around.
819 - cleanup the prototype of data recv and force users to specify when they
820 want to handle references to objects. Test case working even for cycles.
821 - plug memleaks. Valgrind is perfectly ok with this.
823 2004-06-12 Martin Quinson
825 - cleanup the separation between plugin and main code in plugin creation
827 2004-06-11 Martin Quinson
829 - Reput hook for raw sockets, needed for BW experiments
830 - kill a few lines of dead code
831 [Data description] Interface cleanup
832 - gras_datadesc_by_name returns the searched type or NULL.
833 That way, no variable is needed to use a type desc once, which makes
835 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
836 two parameters were strings (field name, type name), leading to
838 [Dicos] Interface cleanup
839 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
840 This is consistant with the dynar API.
842 2004-04-21 Martin Quinson
844 - Porting to new standards.
847 There is no bag anymore, no need to take extra provision to mask the
848 pointers behind "ID".
849 Better splitup of functions between files create/exchange/convert.
850 This is still a bit artificial since convert and receive are so
851 interleaved, but anyway.
853 - add a queued message list to procdata (the ones not matching criteria
855 - factorize some more code between SG and RL wrt procdata
857 - use gras_exit in example to track memleaks
858 - get rid of gs_example now that GS is properly integrated into gras
859 - update run_test to integrate the lastest tests (datadesc)
861 - rename WARNINGn macros to WARNn since it prooved error-prone
863 2004-04-19 Martin Quinson
865 - register init/exit functions within gras module mechanism
866 - send/receive function.
867 Convertion is not implemented, but short-cutted if not needed.
868 struct/array elements are sent one by one (instead of block-wise), but
869 nobody really cares (yet). Get a prototype before optimizing.
870 - tests (using a file socket) for DD send/receive on:
871 - base types: int, float
872 - array: fixed size, string (ie ref to dynamic string)
873 - structure: homogeneous, heterogeneous
874 - chained list, graph with cycle
875 Believe it or not, valgrind is not too unhappy with the results. The
876 cycle happily segfaults, but the others are ok. And I'm sick of pointers
880 - Bugfix when using a filename explicitely (instead of '-')
882 2004-04-09 Martin Quinson
884 - factorize more code between RL and SG in socket creation
885 - Complete the implementation and tests of:
887 o file (only in RL, and mainly for debugging)
889 I lost 3 days to design a portable address resolver, and then decided
890 that the prototype mainly have to run on my box.
891 Addressing portability too early may be like optimizing too early :-/
893 - use gras_init in the Tests instead of the crappy parse_log_opt
894 (the latter function is removed)
895 [Conditional execution]
896 - New functions: gras_if_RL/gras_if_SG (basic support for this)
897 [Code reorganisation]
898 - Get rid of libgrasutils.a since it makes more trouble than it solves.
899 Build examples against the RL library, since there is no way to disable
900 its creation for now.