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]
24 * New! a real RPC mecanism, as it ought to be since too long. [MQ]
25 Exception occuring on server-side are propagated back to client (!).
27 API CHANGE: the callback changed their prototype. Change:
28 int my_handler(gras_socket_t expeditor, void *payload_data) {
30 int my_handler(gras_msg_cb_ctx_t ctx , void *payload_data) {
31 gras_socket_t expeditor=gras_msg_cb_ctx_from(ctx);
36 - Do not force experiment sizes to be expressed in kb, or it becomes
37 impossible to mesure the latency this way (needs one byte-long tests)
38 WARNING: this changes the amok_bw_* function semantic.
42 SimGrid (3.0.1) unstable; urgency=low
45 * Unfortunately, I had missed 5 misnamed functions:
46 xbt_fifo_item_t xbt_fifo_newitem(void);
47 void xbt_fifo_freeitem(xbt_fifo_item_t);
48 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
49 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
50 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
51 They're now deprecated. Please use their new versions:
52 xbt_fifo_item_t xbt_fifo_new_item(void);
53 void xbt_fifo_free_item(xbt_fifo_item_t);
54 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
55 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
56 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
58 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
59 * Documentation: xbt_log module unmercifully reworked [MQ]
60 * Bugfix: there was a problem with the ending of contexts with
61 the pthread backend. It caused some weird deadlock or behavior
62 depending on the pthread implementation. [AL]
63 * Bugfix: get the exceptions raised in the simulator repport where
64 and why they come from when they are not catched in time [AL, MQ]
67 * Bugfix: Do repport the error when two non-connected hosts try to
68 exchange data (Thanks to Flavien for stumbling into this one) [AL]
71 * Add additionnal checkings on communications. Assert that two
72 communicating hosts are connected by a set of links... [AL]
75 * Add additionnal checkings on channel values in communication [AL]
76 * New: MSG_task_get_source to see on which host a task was generated [HC]
77 * New: int MSG_task_probe_from_host(int channel, m_host_t host): returns
78 the number of tasks waiting to be received on channel and sent
80 * New: MSG_error_t MSG_task_get_from_host(m_task_t * task, int channel, m_host_t host);
81 waits for the first task coming from a given host.. [AL]
83 GRAS new functionnalities: [MQ]
84 * Enhance the parsing macro to allow the size of multidimentional objects
85 to be given thru annotations.
86 * New example (and documentation): Matrix Multiplication a la RPC
87 (as when I was young!) and fix a bunch of bugs found on the way.
89 GRAS performance improvements: [MQ]
91 * Reduce the amount of cbps creation/destruction by making it static to
92 datadesc_send/recv() and using a (newly created) cbps_reset (based on
95 * Change libdata to a set so that we can search for stuff by ID (and thus
96 reduce the insane amount of dict lookups)
99 * Actually implement gras_datadesc_copy() so that we don't have to mimick
100 RL communication on top of SG since it's so uneffective.
101 It may also be used for inter-thread communication in RL, one day.
102 * Use gras_datadesc_copy() to exchange messages on top of SG
104 - improve message exchange performance on top of SG
105 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
106 * Don't exchange on the network the size of the used part of buffer,
107 instead, specify the possible buffer size to read().
109 - reduces the amount of read/write calls (one pair per exchange)
110 - reduces the amount of exchanged data (the size)
111 - allows to retrieve all arrived data on receiver side, if we don't need
112 it right now (subsequent read will peek the buffer)
113 - allows the receiver to proceed with the begining of the stream before
114 everything is arrived
115 - make it possible to build an iov transport (using readv/writev)
117 - take care of the data with non-stable storage (like stacked data),
119 * If possible, TCP send uses vector I/O (when writev() is here)
120 - Don't use it for receive since we send data sizes and data on the
121 same stream, so we wouldn't be able to chain large amount of chunks
122 before having to flush the stuff to read the size.
123 * Rework the transport plugin mecanism to simplify it and reduce the
124 amount of pointer dereferencement when searching for the right function
127 * I guess that now, we do almost as few system calls as possible while
128 doing as few data copy as possible.
130 To improve it further, we could try to send all the sizes first and then
131 all the data (to use iov on receiving size), but it's only a partial
132 solution: when you have 2 dimensional data, the sizes of the second
133 dimension is data of the first dimension, so you need 3 streams.
135 I'm not sure the potential performance gains justify the coding burden.
137 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 21 Oct 2005 14:42:20 +0200
139 SimGrid (3.00) stable; urgency=low
142 * New! Give the possibility to hijack the surf parser and thus bypass
143 MSG_create_environment and MSG_launch_application. Have a look at
144 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
146 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
148 SimGrid (2.96) unstable; urgency=low
153 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
154 This deprecates the xbt_error_t mecanisms.
155 It modifies (simplifies) all XBT and GRAS API.
156 MSG API keeps unchanged (exceptions raised by XBT are catched from
157 within MSG and masked with existing error handling facilities)
160 * New! Add a FATPIPE model. [AL]
161 * New! Add a parallel task model. [AL]
162 * New! Add automatically a loopback interface (in the default
163 network model) if none was precised.
166 * Bugfix: MSG_process_resume now works with the current running process.
168 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
169 * Modification of MSG_task_get_compute_duration. Once a task has been
170 processed, the value returned by this function is now equal to 0. [AL]
171 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
172 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
173 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
174 of a m_task. For now, MSG_task_cancel only works with computation
176 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
177 of the processor (in Mflop/s) regardless of the current load on the
179 * API Change: use proper naming convention for MSG_getClock and
180 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
182 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
183 This function changes the priority of a computation task. This priority
184 doesn't affect the transfer rate. A priority of 2 will make a task
185 receive two times more cpu power than the other ones. This function
186 has been added to suit the needs of Nguyen The Loc and hasn't been that
187 much tested yet. So if it fails, please report it and send me your code.
189 * API Change: removed all functions and types that were marked "deprecated"
190 since many months. Renamed MSG_global_init_args to MSG_global_init.
192 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
194 SimGrid (2.95) unstable; urgency=low
197 * Steal some nice code to GNU pth to fix context detection and usage [AL]
198 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
199 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
202 * Bugfix: Allow absolute paths for platform description files [MQ]
203 * Bugfix: do free the variables after use. Leads to drastic performance
205 * Implement max_duration (ie, timeouts) on resources [AL]
208 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
210 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
211 the needed support (timeouts) [AL]
214 * Implement measurement sockets. You can now get the bandwidth between two
215 hosts thanks to AMOK (see below). [MQ]
216 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
217 dynar over the network (yeah) [MQ]
218 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
219 * simple/static token-ring example. [Alexandre Colucci and MQ]
220 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
224 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
225 type parsing. "short" and "long" modifiers were also ignored; other
226 modifier (such as reference level) are still ignored. [MQ]
227 * Embeed the buffer size within the buffer itself on SG. [MQ]
228 That way, send() are atomic and cannot get intermixed anymore (at least
229 the ones which are less than 100k; bigger messages still have the issue)
230 * Array size pushed by the field, not by the field type (or each
231 and every long int will push stuff to the cbps) [MQ]
232 * use select() to sleep since it allows to portably sleep less than one
235 GRAS (minor cleanups)
236 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
239 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
240 * type category 'ignored' killed as it was never used and were difficult
242 * whether a type can cycle or not is now a flag, leaving room for more
243 flags (as "ignored", if we feel the need one day ;)
244 * Rename raw sockets to measurement sockets since "raw" has another
245 meaning in networking community.
248 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
249 and offers features not belonging to GRAS but that most applications
250 need. One day, it may be a set of plugins loadable at runtime.
251 * New module: bandwidth
252 bandwidth measurement between arbitrary nodes running this module. [MQ]
254 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
256 SimGrid (2.94) unstable; urgency=low
258 The first beta release of SimGrid 3 !
262 * Update the main page and the FAQ. Adding references to gforge.
265 * Add a gras_os_getpid function.
268 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
269 * Extend the logs so that they also print PID, hostname, date, ... if
271 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
272 and kill the old version which were in testsuite/
273 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
277 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
278 for users. Instead, each programming environment (GRAS, MSG,...) use it
279 when needed to provide such a feature to users.
280 Don't shortcut the mecanism or you will also shortcut the virtualization
281 you need on the simulator.
285 * Cleanups in configury with regard to compile optimization/warning flags.
286 Also add -fno-loop-optimize to any powerpc since it's the optimization
287 killing gcc (< 3.4.0).
288 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
289 needed by MSG examples complications
290 * Borrow configury beautifications from PHP
293 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
294 warning (thanks loris for stumbling into this one).
295 * Bugfix: stop loading private headers (gras_config.h) from the public
299 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
300 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
301 RL would imply a DNS resolver, which is *hard* to do in a portable way
302 (and therefore delayed).
303 * Implement a real timer mecanism and use it in timing macros. This allows
304 to avoid rounding errors and get a 0.000005 sec precision in timing
305 macros. While I was at it, various cleanups:
306 - allow to declare more than one timed section per file (fix a stupid bug)
307 - move some private declaration to the right place
308 - merge conditional execution and timing macros into emulation module
309 - document the module
310 - make sure the module cleanups its mess on gras_exit
311 * Documentation improvements:
312 - (new) how to compile applications using GRAS
313 - (new) emulation support (timing macros)
315 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
317 SimGrid (2.93) unstable; urgency=low
319 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
322 - Use Paje properly where used. Still to be sanitized properly.
323 - Portability fix: Add an implementation of the contexts using pthread
327 - Add xbt_procname(): returns the name of the current process.
328 Use it to show the current process's name in all logging.
330 - fix detection of older flex version and the reaction, since we do
331 depend on modern ones (we use lex_destroy)
332 - Better separation of SG and RL in the libs: remove all simulation code
333 from libgras. As a result, this lib is now only 200k when stripped.
334 Some of the xbt modules may also be duplicated (two sets and such) and
335 should be cleaned/killed before SG3.
336 - Insist on using xlC on AIX because of weird problems involving gcc there.
337 - Cleanup the make remote stuff. This is now done by scripts
338 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
339 mainly for our private use, but we're working on changing them to user
342 - Bugfix: flush the socket on close only if there is some *output*.
343 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
345 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
348 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
350 SimGrid (2.92) unstable; urgency=low
352 Alpha 3 on the path to SimGrid 3
356 - New! First try of benchmarking macros.
357 - New! First try so that gras_stub_generator generate deployment and
358 remote compilation helpers.
360 - Bugfix: Initialization fix in msg_test.
364 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
366 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
368 - API Change: gras_os_sleep to take the amount of seconds as a double.
369 Accepting an int was error prone since it was the only location where
370 seconds were coded as such. It leaded to damn rounding errors.
371 - Bugfix: Hard to belive that timers ever worked before this.
373 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
375 SimGrid (2.91) unstable; urgency=low
377 Alpha 2 on the path to SimGrid 3
381 - Bug fix in the lmm_solver.
383 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
384 through the function MSG_paje_output.
385 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
386 - It is possible to bound the rate of a communication in MSG with
387 MSG_task_put_bounded() (was already in the previous version but I had forgotten
388 to write it in the changelog).
389 - Bug fix to let GRAS run on top of MSG until we move it directly on top
394 - Various cleanups to the autotools stuff
395 - Begin to move Gras examples to examples/gras/
396 - Let make distcheck work again (yeah!)
398 - documentation overhauled using doxygen.
399 gtk-doc-tools is dead in SimGrid now.
400 - Automatically extract all existing logging categories, and add the list
401 to the documentation (long standing one, to say the less)
403 - Cleanup the known architecture table. Reorder the entries to group what
404 should be, and use a more consistent naming scheme.
405 (some of the test dataset are still to be regenerated)
406 - New! Allow library to register globals on each process just as userdata
408 This is implemented using a xbt_dict and not a xbt_set, so we loose the
409 lookup time (for now).
410 Use it in msg and trp.
411 This cleans a lot the internals and helps enforcing privacy of the
412 headers between the gras components.
413 - New! Add a timer mechanism, not unlike cron(8) and at(1).
414 - Bugfix: gras_os_time was delirious in RL.
415 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
417 - Reenable GRAS now that it works.
419 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
421 SimGrid (2.90) unstable; urgency=low
423 Alpha 1 on the path to SimGrid 3
425 * It is a long time since the last release of SimGrid. I'm sorry about
426 that but as I had told you, I was rewriting a lot of things. I apologize
427 to those who had been reporting bugs to me and that I had not answered.
428 If your bug is still in the new version, please tell me. Here is a
429 summary of the main changes.
431 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
432 lead by Martin Quinson. As a consequence SimGrid gains a lot in
433 portability, speed, and a lot more but you'll figure it out later.
434 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
435 I wanted to release the new MSG as soon as possible and I have
436 broken GRAS, which is the reason why, for now, only MSG is fully
437 functional. A laconic description of these projects is available
438 in the documentation.
440 * REVOLUTION 2: I have removed SG and I am now using a new simulation
441 kernel optimized for our needs (called SURF but only the developers
442 should use it). Hence, MSG is now roughly 30 times faster and I think
443 that by rewriting a little bit MSG, I could event speed it up a little
444 bit more. Beside the gain in speed, it is also much easier to encode a
445 new platform model with SURF than it was with SG. More to come...
447 * REVOLUTION 3: I have tried to change a little as possible the API of
448 MSG but a few things really had to disappear. The main differences
449 with the previous version are :
450 1) no more m_links_t and the corresponding functions. Platforms are
451 directly read from a XML description and cannot be hard-coded
452 anymore. The same format is used for application deployment
453 description. The new format is described in the documentation.
454 Have a look in tools/platform_generation. There is a tiny script
455 that converts from the old platform format to the new one. Concerning
456 the application deployment format, parsing the old one is tricky.
457 I think most of you should however be able to convert your files. If
458 it is really an issue, I can write a C code that does the conversion.
460 2) the toolbox tbx does not exist anymore. We now have a library
461 with much more data-structures but without the hash-tables (we have
462 dictionaries that are much faster).
464 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
466 *****************************************************************************
467 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
468 * don't feel like converting the oldies. (Mt) *
469 *****************************************************************************
472 Version 2.90: "the long awaited one"
473 - Finished rewriting and debugging MSG. Rewrote the documentation.
474 - disable GRAS for now since it needs to be ported to the newest SG
477 - Finish the port to windows (using mingw32 for cross-compile)
480 - Main loop and datastructures of SURF. A cpu resource object is
481 functional. Surf can thus be used to create cpu's with variable
482 performance on which you can execute some actions.
484 2004-11-15 Martin Quinson
485 - Port to ARM. Simply added the alignment and size descriptions. Should
486 work, but the ARM machines are so slow that I didn't had the opportunity
487 to 'make check' over there yet.
489 2004-11-15 Arnaud Legrand
490 - Trace manager now written. It uses a heap structure and is therefore
491 expected to be efficient. It may however be speeded up (particularly
492 when many events occur at the same date) by using red and black
493 trees. One day maybe...
494 - Max-min linear system solver written. It uses a sparse matrix
495 structure taking advantage of its expected use. Most operations are
496 O(1) and free/calloc are called as few as possible. The computation of
497 the minimum could however be improved by using a red and black tree
500 2004-11-03 Arnaud Legrand
501 - Rename every gras_* function that was in xbt/ to its xbt_
503 - Add a heap and a doubly-linked list to xbt
504 - Added a dichotomy to the dictionaries. make check works as well before
505 so I assume that the patch is correct. I do not know however if things
506 run effectively faster than before now. :)
508 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
509 SimGrid, and the version number is bumped to 2.x
511 2004-10-29 Martin Quinson
512 - Introduction of the remote errors.
513 They are the result of a RMI/RPC on the remote machine.
514 ErrCodes being scalar values, you can't get the host on which those
515 errors did happen. Extending the error mechanism as in Gnome is possible.
516 No idea yet whether it is a good idea.
518 2004-10-28 Martin Quinson
519 - Interface revolution: the Starred Structure Eradication.
520 I used to do typedef struct {} toto_t; and then handle *toto_t.
521 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
522 - ???_t is a valid type (builded with typedef)
523 - s_toto_t is a structure (access to fields with .)
524 - s_toto is a structure needing 'struct' keyword to be used
525 - e_toto_t is an enum
526 - toto_t is an 'object' (struct*)
528 typedef struct s_toto {} s_toto_t, *toto_t;
529 typedef enum {} e_toto_t;
530 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
531 s_toto_t) is private.
533 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
534 so that it can set the variable to NULL. It was so for dicts and sets,
535 it changed for dynars.
537 - Fix a bunch of memleaks in dict_remove
538 - Fix a bug in sg/server_socket opening: it failed all the time.
540 2004-10-07 Martin Quinson
541 - Speed up dynar lookup operation a bit.
543 gras_dynar_get is dead.
545 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
546 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
547 (which gives you the address of the stored data).
549 gras_dynar_get_as is an helpful macro which allows you to retrieve a
550 copy of the data using an affectation to do the job and not a memcpy.
552 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
553 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
555 It does not really speedup the dynar test because they are
556 setting elements all the time (and look them seldom). But the dict does
557 far more lookup than setting.
559 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
561 2004-10-05 Martin Quinson
562 - Allow to (en/dis)able the cycle detection at run time.
564 Whether we should check for cycle or not is now a property of each
565 datatype. When you think there may be some cycle, use datadesc_cycle_set.
566 datadesc_cycle_unset allow to remove this property when previously set.
568 Note that the cycle detection is off by default since it impacts the
569 performance. Watch the data you feed GRAS with ;)
571 This property is hereditary. Any element embedded in a structure having it
572 set have it set for the time of this data exchange.
574 You should set it both on sender and receiver side. If you don't set it on
575 sender side, it will enter an endless loop. If you forget on receiver
576 side, the cycles won't be recreated after communication.
578 - Header reorganization.
579 Kill gras_private.h, each submodule must load the headers it needs.
581 2004-10-04 Martin Quinson
582 - Interface revolution: do not try to survive to malloc failure.
584 Now, gras_malloc and friends call gras_abort() on failure.
585 As a conclusion, malloc_error is not a valid error anymore, and all
586 functions for which it was the only gras_error_t return value are
587 changed. They now return void, or there result directly.
588 This simplify the API a lot.
590 2004-09-29 Martin Quinson
591 - Re-enable raw sockets.
592 Created by gras_socket_{client,server}_ext;
593 Used with gras_raw_{send,recv}
596 It should allow to kill the last bits of gras first version soon.
598 This is not completely satisfactory yet (duplicate code with
599 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
602 - Simplify transport plugin (internal) interface by not passing any
603 argument to _server and _client, but embedding them in the socket
606 2004-09-28 Martin Quinson
607 - Finish the port to AIX.
608 autoconf was my problem (segfault within the malloc replacement
609 function. No idea why)
611 2004-09-16 Martin Quinson
612 - Fix some size_t madness on 64bit architectures.
614 2004-09-08 Martin Quinson
615 - Reduce the number of system headers loaded, overload some more system
616 calls (such as malloc to cast the result of the system one, and work
618 - Fix and reintroduce the config support
620 2004-09-07 Martin Quinson
621 - Source code reorganization to allow Arnaud to surf all over there.
622 - Allow to document the logging categories.
623 - Remove all uppercase from logging categories and useless cleanup in names.
625 2004-08-18 Martin Quinson
626 Version 0.6.2 (protocol not changed; API changed)
627 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
628 gras_error_t), and NULL when not found. Functions expecting a msgtype
629 as argument (msg_wait; msg_send) deal with NULL argument by providing a
630 hopefully usefull message.
631 - Portability to prehistoric sparcs again
633 2004-08-17 Martin Quinson
634 Version 0.6.1 (protocol not changed; ABI not changed)
635 - prealloc some buffers to speed things up
637 2004-08-11 Martin Quinson
638 Version 0.6 (protocol not changed; ABI expended)
639 - The parsing macro can deal with the references, provided that you add
640 the relevant annotations (using GRAS_ANNOTE(size,field_name))
642 2004-08-09 Martin Quinson
643 Version 0.5 (protocol not changed; ABI changed)
644 - Allow to off turn the cycle detection code in data exchange at
645 compilation time. It should be at run time, but I'm short of time (and
646 the config stuff is still broken). That way, we keep dict out of the
647 critical path, which is good because the performance is poor:
648 - search not dichotomial yet
649 - dynar give no way to access their content and memcpy everytime
650 - In composed data description (struct, ref and so on), stop foolness of
651 keeping the subtype's ID, but store the type itself. This keeps sets out
652 of the critical path, which is good since they rely on dynar and
653 dictionnaries. The only loose of that is that we cannot detect the
654 redeclaration of a structure/union with another content (but I'm not sure
655 the code detected well this error before anyway). We still can detect
656 the redefinition discrepancy for the other types.
657 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
658 it breaks the code because of type-punning used all over the place).
659 This breaks on all non-gcc architectures (for now).
661 All those changes (plus the buffer of last time) allow me to gain 2 order
662 of magnitude on cruel tests consisting of 800000 array of integers on two
663 level of a hierarchical structure (200 secondes -> 4 secondes)
666 - the selector of reference must now return the type it points to, not
669 2004-08-06 Martin Quinson
670 Version 0.4 (protocol changed; ABI not changed)
671 - Allow to pass --gras-log argument to processes in simulation mode. Really.
672 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
673 - Implement a buffer transport, and use it by default (it relies on tcp in
674 real life and on sg in simulation).
675 That's a bit hackish since I had a new field to the structure to store
676 its data without interfering with the subtype ones. Inheritance
677 is tricky in C. And that's a kind of reverse inheritance with one class
678 derivating two classes. Or maybe a game with java interfaces. Anyway,
679 that's damn hard in C (at least).
680 Moreover, I got tired while trying to ensure plugin separation and
681 genericity in SG mode. MSG wants me to do weird things, so let's go for
682 cruel hacks (temporarily of course ;).
683 See comment in transport_private.h:71
684 - do not include all the _interface headers in private but in the files
685 which really need them (to cut the compilation time when they are
688 2004-07-26 Martin Quinson
689 Version 0.3 (protocol not changed; ABI changed)
690 - Major overhault of the datadesc interface to simplify it:
691 - shorted the function names:
692 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
693 - add a trivial way to push/pop integers into the cbps without malloc.
694 This allows to make really generic sub_type description, which simply
695 pop their size of the stack.
696 - add a function gras_datadesc_ref_pop_arr() which does what users want
697 most of the time: Declare a dynamic array (which pops its size of the
698 stack) and declare a reference to it. Poor name, but anyway.
699 - kill the post-send callback, add a post-receive one
701 2004-07-23 Martin Quinson
702 Version 0.2 (protocol changed; ABI changed)
703 - add some testing for cpbs in the test cases, and fix some more bugs.
704 This invalidate again the little64 data file, since I cannot regenerate
706 - remove an awfull optimization in the logging stuff, allowing me to:
707 - understand it again
708 - learn gcc how to check that the argument match the provided format
709 - fix all errors revealed by gcc after that
710 - internal keys of dict are not \0 terminated. Deal with it properly in
711 loggings instead of segfaulting when the user want to see the logs :-/
713 2004-07-22 Martin Quinson
714 - Fix some stupid bug preventing cbps (callback postit) from working
716 2004-07-21 Martin Quinson
717 - Some documentation cleanups
718 - remove the useless last argument of msgtype_declare
719 - rename the Virtu functions to fit into the 'os' namespace
720 - move headers src/include -> src/include/gras/ and stop fooling with
721 gras -> . symbolic link
722 - make distcheck is now successful
724 2004-07-19 Martin Quinson
726 - Build shared library also
727 - Install html doc to the right location
728 - stop removing maintainer files in make clean
729 - build tests only on make check
731 2004-07-13 Martin Quinson
733 - No major issue in previous version => change versionning schema
734 - Re-enable little64 convertion test now that Abdou kindly regenerated the
735 corresponding dataset.
737 2004-07-11 Martin Quinson
739 - Get it working with any kind of structure (we can compute the padding
740 bytes remotely for all the architectures I have access to)
741 - Implement the structure parsing macro (still not quite robust/complete)
742 - Improvement to the remote testing toysuite
744 2004-07-10 Martin Quinson
746 - get ride of a bunch of deprecated macros
747 - actually run the test for two-compliment, not only compile it :-/
748 - test whether the structures get packed (and bail out if yes. Damn.
749 Alignment is a serious matter)
750 - test whether the structures get compacted (but respecting the alignment
751 constraints of each types)
752 - test whether the array fields of structures can straddle alignment boundaries
754 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
755 (took me 2 hours to find that bug, looking at the wrong place)
756 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
757 before sending/receiving and are used to compute the offsets of fields
758 - Given that padding size depend even on compiler options, keep track of
759 alignment and aligned_size only for the current architecture. Not a big
760 deal since we send structure fields one after the other (seems
762 - Add the datastructure used for IEEE paper by the PBIO guys to the test
763 program, let it work on linux/gcc/little32. portability todo.
765 2004-07-08 Martin Quinson
766 - import and improve remote compilation support from FAST
767 - make sure make check works on half a dozen of machines out there
769 2004-07-07 Martin Quinson
770 Let's say it's version 0.0.3 ;)
771 - Implement conversions (yuhu!)
772 - Let it work on solaris (beside conversion, of course)
773 - Stupid me, using rand() to generate the conversion datatests in not wise.
775 2004-07-06 Martin Quinson
776 - Let make dist work, since I'm gonna need it to compile on remote hosts
777 - Let Tests/datadesc_usage write the architecture on which the file was
778 generated as first byte.
779 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
780 architecture descriptions.
781 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
782 on those architectures.
783 - Optimization: send/recv array of scalar in one shoot
785 2004-07-05 Martin Quinson
786 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
788 - Plug a whole bunch of memleaks
789 - each process now have to call gras_{init,exit}. One day, their log
790 settings will be separated
791 - Continue the code factorisation between SG, RL and common in Transport.
793 2004-07-04 Martin Quinson
795 - Redistribution between SG and RL.
796 We wanna have to accept in SG, so move accepted related parts of RL in
797 the common part. (more precisely, the dynar of all known sockets is no
798 more a static in transport.c, but part of the process_data)
800 [gras_stub_generator]
801 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
803 2004-07-03 Martin Quinson
804 - Create a new log channel tbx containing dict, set, log, dynar (to shut
805 them all up in one shot)
807 - Fix the ugly case of reference to dynamic array.
808 - New (semi-public) function gras_datadesc_size to allow the messaging
809 layer to malloc the needed space for the buffer.
811 - gras_socket_close now expect the socket to close (and not its address to
812 put NULL in it after it). This is because the socket passed to handlers
813 is one of their argument (=> not writable).
815 - propagate the interface cleanup from last week in datadesc, ie remove a
816 superfluous level of indirection. User pass adress of variable
817 containing data (both when sending and receiving), and not of a variable
818 being a pointer to the data. Let's say that I like it better ;)
819 The price for that is constructs like "int msg=*(int*)payload" in
820 handlers, but it's a fine price, IMHO.
822 - Let it work in RL (yuhu)
824 2004-06-21 Martin Quinson
826 - porting SG plugin and SG select to new standards (works almost).
827 - plug memleaks and fix bugs around.
830 - cleanup the prototype of data recv and force users to specify when they
831 want to handle references to objects. Test case working even for cycles.
832 - plug memleaks. Valgrind is perfectly ok with this.
834 2004-06-12 Martin Quinson
836 - cleanup the separation between plugin and main code in plugin creation
838 2004-06-11 Martin Quinson
840 - Reput hook for raw sockets, needed for BW experiments
841 - kill a few lines of dead code
842 [Data description] Interface cleanup
843 - gras_datadesc_by_name returns the searched type or NULL.
844 That way, no variable is needed to use a type desc once, which makes
846 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
847 two parameters were strings (field name, type name), leading to
849 [Dicos] Interface cleanup
850 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
851 This is consistant with the dynar API.
853 2004-04-21 Martin Quinson
855 - Porting to new standards.
858 There is no bag anymore, no need to take extra provision to mask the
859 pointers behind "ID".
860 Better splitup of functions between files create/exchange/convert.
861 This is still a bit artificial since convert and receive are so
862 interleaved, but anyway.
864 - add a queued message list to procdata (the ones not matching criteria
866 - factorize some more code between SG and RL wrt procdata
868 - use gras_exit in example to track memleaks
869 - get rid of gs_example now that GS is properly integrated into gras
870 - update run_test to integrate the lastest tests (datadesc)
872 - rename WARNINGn macros to WARNn since it prooved error-prone
874 2004-04-19 Martin Quinson
876 - register init/exit functions within gras module mechanism
877 - send/receive function.
878 Convertion is not implemented, but short-cutted if not needed.
879 struct/array elements are sent one by one (instead of block-wise), but
880 nobody really cares (yet). Get a prototype before optimizing.
881 - tests (using a file socket) for DD send/receive on:
882 - base types: int, float
883 - array: fixed size, string (ie ref to dynamic string)
884 - structure: homogeneous, heterogeneous
885 - chained list, graph with cycle
886 Believe it or not, valgrind is not too unhappy with the results. The
887 cycle happily segfaults, but the others are ok. And I'm sick of pointers
891 - Bugfix when using a filename explicitely (instead of '-')
893 2004-04-09 Martin Quinson
895 - factorize more code between RL and SG in socket creation
896 - Complete the implementation and tests of:
898 o file (only in RL, and mainly for debugging)
900 I lost 3 days to design a portable address resolver, and then decided
901 that the prototype mainly have to run on my box.
902 Addressing portability too early may be like optimizing too early :-/
904 - use gras_init in the Tests instead of the crappy parse_log_opt
905 (the latter function is removed)
906 [Conditional execution]
907 - New functions: gras_if_RL/gras_if_SG (basic support for this)
908 [Code reorganisation]
909 - Get rid of libgrasutils.a since it makes more trouble than it solves.
910 Build examples against the RL library, since there is no way to disable
911 its creation for now.