1 SimGrid (3.0.1) unstable; urgency=low
4 * Unfortunately, I had missed 5 misnamed functions:
5 xbt_fifo_item_t xbt_fifo_newitem(void);
6 void xbt_fifo_freeitem(xbt_fifo_item_t);
7 xbt_fifo_item_t xbt_fifo_getFirstItem(xbt_fifo_t l);
8 xbt_fifo_item_t xbt_fifo_getNextItem(xbt_fifo_item_t i);
9 xbt_fifo_item_t xbt_fifo_getPrevItem(xbt_fifo_item_t i);
10 They're now deprecated. Please use their new versions:
11 xbt_fifo_item_t xbt_fifo_new_item(void);
12 void xbt_fifo_free_item(xbt_fifo_item_t);
13 xbt_fifo_item_t xbt_fifo_get_first_item(xbt_fifo_t l);
14 xbt_fifo_item_t xbt_fifo_get_next_item(xbt_fifo_item_t i);
15 xbt_fifo_item_t xbt_fifo_get_prev_item(xbt_fifo_item_t i);
17 * Bugfix: really disconnect fifo items which are remove_item()ed [AL]
20 * Add addtionnal checkings on channel values in communicating functions.
23 * Actually implement gras_datadesc_copy() so that we don't have to mimick
24 RL communication on top of SG since it's so uneffective.
25 It may also be used for inter-thread communication in RL, one day. [MQ]
27 Documentation improvements:
28 * xbt_log module unmercifully reworked [MQ]
32 SimGrid (3.00) stable; urgency=low
35 * New! Give the possibility to hijack the surf parser and thus bypass
36 MSG_create_environment and MSG_launch_application. Have a look at
37 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
39 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
41 SimGrid (2.96) unstable; urgency=low
46 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
47 This deprecates the xbt_error_t mecanisms.
48 It modifies (simplifies) all XBT and GRAS API.
49 MSG API keeps unchanged (exceptions raised by XBT are catched from
50 within MSG and masked with existing error handling facilities)
53 * New! Add a FATPIPE model. [AL]
54 * New! Add a parallel task model. [AL]
55 * New! Add automatically a loopback interface (in the default
56 network model) if none was precised.
59 * Bugfix: MSG_process_resume now works with the current running process.
61 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
62 * Modification of MSG_task_get_compute_duration. Once a task has been
63 processed, the value returned by this function is now equal to 0. [AL]
64 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
65 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
66 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
67 of a m_task. For now, MSG_task_cancel only works with computation
69 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
70 of the processor (in Mflop/s) regardless of the current load on the
72 * API Change: use proper naming convention for MSG_getClock and
73 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
75 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
76 This function changes the priority of a computation task. This priority
77 doesn't affect the transfer rate. A priority of 2 will make a task
78 receive two times more cpu power than the other ones. This function
79 has been added to suit the needs of Nguyen The Loc and hasn't been that
80 much tested yet. So if it fails, please report it and send me your code.
82 * API Change: removed all functions and types that were marked "deprecated"
83 since many months. Renamed MSG_global_init_args to MSG_global_init.
85 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
87 SimGrid (2.95) unstable; urgency=low
90 * Steal some nice code to GNU pth to fix context detection and usage [AL]
91 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
92 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
95 * Bugfix: Allow absolute paths for platform description files [MQ]
96 * Bugfix: do free the variables after use. Leads to drastic performance
98 * Implement max_duration (ie, timeouts) on resources [AL]
101 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
103 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
104 the needed support (timeouts) [AL]
107 * Implement measurement sockets. You can now get the bandwidth between two
108 hosts thanks to AMOK (see below). [MQ]
109 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
110 dynar over the network (yeah) [MQ]
111 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
112 * simple/static token-ring example. [Alexandre Colucci and MQ]
113 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
117 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
118 type parsing. "short" and "long" modifiers were also ignored; other
119 modifier (such as reference level) are still ignored. [MQ]
120 * Embeed the buffer size within the buffer itself on SG. [MQ]
121 That way, send() are atomic and cannot get intermixed anymore (at least
122 the ones which are less than 100k; bigger messages still have the issue)
123 * Array size pushed by the field, not by the field type (or each
124 and every long int will push stuff to the cbps) [MQ]
125 * use select() to sleep since it allows to portably sleep less than one
128 GRAS (minor cleanups)
129 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
132 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
133 * type category 'ignored' killed as it was never used and were difficult
135 * whether a type can cycle or not is now a flag, leaving room for more
136 flags (as "ignored", if we feel the need one day ;)
137 * Rename raw sockets to measurement sockets since "raw" has another
138 meaning in networking community.
141 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
142 and offers features not belonging to GRAS but that most applications
143 need. One day, it may be a set of plugins loadable at runtime.
144 * New module: bandwidth
145 bandwidth measurement between arbitrary nodes running this module. [MQ]
147 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
149 SimGrid (2.94) unstable; urgency=low
151 The first beta release of SimGrid 3 !
155 * Update the main page and the FAQ. Adding references to gforge.
158 * Add a gras_os_getpid function.
161 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
162 * Extend the logs so that they also print PID, hostname, date, ... if
164 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
165 and kill the old version which were in testsuite/
166 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
170 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
171 for users. Instead, each programming environment (GRAS, MSG,...) use it
172 when needed to provide such a feature to users.
173 Don't shortcut the mecanism or you will also shortcut the virtualization
174 you need on the simulator.
178 * Cleanups in configury with regard to compile optimization/warning flags.
179 Also add -fno-loop-optimize to any powerpc since it's the optimization
180 killing gcc (< 3.4.0).
181 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
182 needed by MSG examples complications
183 * Borrow configury beautifications from PHP
186 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
187 warning (thanks loris for stumbling into this one).
188 * Bugfix: stop loading private headers (gras_config.h) from the public
192 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
193 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
194 RL would imply a DNS resolver, which is *hard* to do in a portable way
195 (and therefore delayed).
196 * Implement a real timer mecanism and use it in timing macros. This allows
197 to avoid rounding errors and get a 0.000005 sec precision in timing
198 macros. While I was at it, various cleanups:
199 - allow to declare more than one timed section per file (fix a stupid bug)
200 - move some private declaration to the right place
201 - merge conditional execution and timing macros into emulation module
202 - document the module
203 - make sure the module cleanups its mess on gras_exit
204 * Documentation improvements:
205 - (new) how to compile applications using GRAS
206 - (new) emulation support (timing macros)
208 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
210 SimGrid (2.93) unstable; urgency=low
212 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
215 - Use Paje properly where used. Still to be sanitized properly.
216 - Portability fix: Add an implementation of the contexts using pthread
220 - Add xbt_procname(): returns the name of the current process.
221 Use it to show the current process's name in all logging.
223 - fix detection of older flex version and the reaction, since we do
224 depend on modern ones (we use lex_destroy)
225 - Better separation of SG and RL in the libs: remove all simulation code
226 from libgras. As a result, this lib is now only 200k when stripped.
227 Some of the xbt modules may also be duplicated (two sets and such) and
228 should be cleaned/killed before SG3.
229 - Insist on using xlC on AIX because of weird problems involving gcc there.
230 - Cleanup the make remote stuff. This is now done by scripts
231 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
232 mainly for our private use, but we're working on changing them to user
235 - Bugfix: flush the socket on close only if there is some *output*.
236 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
238 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
241 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
243 SimGrid (2.92) unstable; urgency=low
245 Alpha 3 on the path to SimGrid 3
249 - New! First try of benchmarking macros.
250 - New! First try so that gras_stub_generator generate deployment and
251 remote compilation helpers.
253 - Bugfix: Initialization fix in msg_test.
257 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
259 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
261 - API Change: gras_os_sleep to take the amount of seconds as a double.
262 Accepting an int was error prone since it was the only location where
263 seconds were coded as such. It leaded to damn rounding errors.
264 - Bugfix: Hard to belive that timers ever worked before this.
266 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
268 SimGrid (2.91) unstable; urgency=low
270 Alpha 2 on the path to SimGrid 3
274 - Bug fix in the lmm_solver.
276 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
277 through the function MSG_paje_output.
278 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
279 - It is possible to bound the rate of a communication in MSG with
280 MSG_task_put_bounded() (was already in the previous version but I had forgotten
281 to write it in the changelog).
282 - Bug fix to let GRAS run on top of MSG until we move it directly on top
287 - Various cleanups to the autotools stuff
288 - Begin to move Gras examples to examples/gras/
289 - Let make distcheck work again (yeah!)
291 - documentation overhauled using doxygen.
292 gtk-doc-tools is dead in SimGrid now.
293 - Automatically extract all existing logging categories, and add the list
294 to the documentation (long standing one, to say the less)
296 - Cleanup the known architecture table. Reorder the entries to group what
297 should be, and use a more consistent naming scheme.
298 (some of the test dataset are still to be regenerated)
299 - New! Allow library to register globals on each process just as userdata
301 This is implemented using a xbt_dict and not a xbt_set, so we loose the
302 lookup time (for now).
303 Use it in msg and trp.
304 This cleans a lot the internals and helps enforcing privacy of the
305 headers between the gras components.
306 - New! Add a timer mechanism, not unlike cron(8) and at(1).
307 - Bugfix: gras_os_time was delirious in RL.
308 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
310 - Reenable GRAS now that it works.
312 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
314 SimGrid (2.90) unstable; urgency=low
316 Alpha 1 on the path to SimGrid 3
318 * It is a long time since the last release of SimGrid. I'm sorry about
319 that but as I had told you, I was rewriting a lot of things. I apologize
320 to those who had been reporting bugs to me and that I had not answered.
321 If your bug is still in the new version, please tell me. Here is a
322 summary of the main changes.
324 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
325 lead by Martin Quinson. As a consequence SimGrid gains a lot in
326 portability, speed, and a lot more but you'll figure it out later.
327 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
328 I wanted to release the new MSG as soon as possible and I have
329 broken GRAS, which is the reason why, for now, only MSG is fully
330 functional. A laconic description of these projects is available
331 in the documentation.
333 * REVOLUTION 2: I have removed SG and I am now using a new simulation
334 kernel optimized for our needs (called SURF but only the developers
335 should use it). Hence, MSG is now roughly 30 times faster and I think
336 that by rewriting a little bit MSG, I could event speed it up a little
337 bit more. Beside the gain in speed, it is also much easier to encode a
338 new platform model with SURF than it was with SG. More to come...
340 * REVOLUTION 3: I have tried to change a little as possible the API of
341 MSG but a few things really had to disappear. The main differences
342 with the previous version are :
343 1) no more m_links_t and the corresponding functions. Platforms are
344 directly read from a XML description and cannot be hard-coded
345 anymore. The same format is used for application deployment
346 description. The new format is described in the documentation.
347 Have a look in tools/platform_generation. There is a tiny script
348 that converts from the old platform format to the new one. Concerning
349 the application deployment format, parsing the old one is tricky.
350 I think most of you should however be able to convert your files. If
351 it is really an issue, I can write a C code that does the conversion.
353 2) the toolbox tbx does not exist anymore. We now have a library
354 with much more data-structures but without the hash-tables (we have
355 dictionaries that are much faster).
357 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
359 *****************************************************************************
360 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
361 * don't feel like converting the oldies. (Mt) *
362 *****************************************************************************
365 Version 2.90: "the long awaited one"
366 - Finished rewriting and debugging MSG. Rewrote the documentation.
367 - disable GRAS for now since it needs to be ported to the newest SG
370 - Finish the port to windows (using mingw32 for cross-compile)
373 - Main loop and datastructures of SURF. A cpu resource object is
374 functional. Surf can thus be used to create cpu's with variable
375 performance on which you can execute some actions.
377 2004-11-15 Martin Quinson
378 - Port to ARM. Simply added the alignment and size descriptions. Should
379 work, but the ARM machines are so slow that I didn't had the opportunity
380 to 'make check' over there yet.
382 2004-11-15 Arnaud Legrand
383 - Trace manager now written. It uses a heap structure and is therefore
384 expected to be efficient. It may however be speeded up (particularly
385 when many events occur at the same date) by using red and black
386 trees. One day maybe...
387 - Max-min linear system solver written. It uses a sparse matrix
388 structure taking advantage of its expected use. Most operations are
389 O(1) and free/calloc are called as few as possible. The computation of
390 the minimum could however be improved by using a red and black tree
393 2004-11-03 Arnaud Legrand
394 - Rename every gras_* function that was in xbt/ to its xbt_
396 - Add a heap and a doubly-linked list to xbt
397 - Added a dichotomy to the dictionaries. make check works as well before
398 so I assume that the patch is correct. I do not know however if things
399 run effectively faster than before now. :)
401 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
402 SimGrid, and the version number is bumped to 2.x
404 2004-10-29 Martin Quinson
405 - Introduction of the remote errors.
406 They are the result of a RMI/RPC on the remote machine.
407 ErrCodes being scalar values, you can't get the host on which those
408 errors did happen. Extending the error mechanism as in Gnome is possible.
409 No idea yet whether it is a good idea.
411 2004-10-28 Martin Quinson
412 - Interface revolution: the Starred Structure Eradication.
413 I used to do typedef struct {} toto_t; and then handle *toto_t.
414 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
415 - ???_t is a valid type (builded with typedef)
416 - s_toto_t is a structure (access to fields with .)
417 - s_toto is a structure needing 'struct' keyword to be used
418 - e_toto_t is an enum
419 - toto_t is an 'object' (struct*)
421 typedef struct s_toto {} s_toto_t, *toto_t;
422 typedef enum {} e_toto_t;
423 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
424 s_toto_t) is private.
426 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
427 so that it can set the variable to NULL. It was so for dicts and sets,
428 it changed for dynars.
430 - Fix a bunch of memleaks in dict_remove
431 - Fix a bug in sg/server_socket opening: it failed all the time.
433 2004-10-07 Martin Quinson
434 - Speed up dynar lookup operation a bit.
436 gras_dynar_get is dead.
438 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
439 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
440 (which gives you the address of the stored data).
442 gras_dynar_get_as is an helpful macro which allows you to retrieve a
443 copy of the data using an affectation to do the job and not a memcpy.
445 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
446 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
448 It does not really speedup the dynar test because they are
449 setting elements all the time (and look them seldom). But the dict does
450 far more lookup than setting.
452 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
454 2004-10-05 Martin Quinson
455 - Allow to (en/dis)able the cycle detection at run time.
457 Whether we should check for cycle or not is now a property of each
458 datatype. When you think there may be some cycle, use datadesc_cycle_set.
459 datadesc_cycle_unset allow to remove this property when previously set.
461 Note that the cycle detection is off by default since it impacts the
462 performance. Watch the data you feed GRAS with ;)
464 This property is hereditary. Any element embedded in a structure having it
465 set have it set for the time of this data exchange.
467 You should set it both on sender and receiver side. If you don't set it on
468 sender side, it will enter an endless loop. If you forget on receiver
469 side, the cycles won't be recreated after communication.
471 - Header reorganization.
472 Kill gras_private.h, each submodule must load the headers it needs.
474 2004-10-04 Martin Quinson
475 - Interface revolution: do not try to survive to malloc failure.
477 Now, gras_malloc and friends call gras_abort() on failure.
478 As a conclusion, malloc_error is not a valid error anymore, and all
479 functions for which it was the only gras_error_t return value are
480 changed. They now return void, or there result directly.
481 This simplify the API a lot.
483 2004-09-29 Martin Quinson
484 - Re-enable raw sockets.
485 Created by gras_socket_{client,server}_ext;
486 Used with gras_raw_{send,recv}
489 It should allow to kill the last bits of gras first version soon.
491 This is not completely satisfactory yet (duplicate code with
492 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
495 - Simplify transport plugin (internal) interface by not passing any
496 argument to _server and _client, but embedding them in the socket
499 2004-09-28 Martin Quinson
500 - Finish the port to AIX.
501 autoconf was my problem (segfault within the malloc replacement
502 function. No idea why)
504 2004-09-16 Martin Quinson
505 - Fix some size_t madness on 64bit architectures.
507 2004-09-08 Martin Quinson
508 - Reduce the number of system headers loaded, overload some more system
509 calls (such as malloc to cast the result of the system one, and work
511 - Fix and reintroduce the config support
513 2004-09-07 Martin Quinson
514 - Source code reorganization to allow Arnaud to surf all over there.
515 - Allow to document the logging categories.
516 - Remove all uppercase from logging categories and useless cleanup in names.
518 2004-08-18 Martin Quinson
519 Version 0.6.2 (protocol not changed; API changed)
520 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
521 gras_error_t), and NULL when not found. Functions expecting a msgtype
522 as argument (msg_wait; msg_send) deal with NULL argument by providing a
523 hopefully usefull message.
524 - Portability to prehistoric sparcs again
526 2004-08-17 Martin Quinson
527 Version 0.6.1 (protocol not changed; ABI not changed)
528 - prealloc some buffers to speed things up
530 2004-08-11 Martin Quinson
531 Version 0.6 (protocol not changed; ABI expended)
532 - The parsing macro can deal with the references, provided that you add
533 the relevant annotations (using GRAS_ANNOTE(size,field_name))
535 2004-08-09 Martin Quinson
536 Version 0.5 (protocol not changed; ABI changed)
537 - Allow to off turn the cycle detection code in data exchange at
538 compilation time. It should be at run time, but I'm short of time (and
539 the config stuff is still broken). That way, we keep dict out of the
540 critical path, which is good because the performance is poor:
541 - search not dichotomial yet
542 - dynar give no way to access their content and memcpy everytime
543 - In composed data description (struct, ref and so on), stop foolness of
544 keeping the subtype's ID, but store the type itself. This keeps sets out
545 of the critical path, which is good since they rely on dynar and
546 dictionnaries. The only loose of that is that we cannot detect the
547 redeclaration of a structure/union with another content (but I'm not sure
548 the code detected well this error before anyway). We still can detect
549 the redefinition discrepancy for the other types.
550 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
551 it breaks the code because of type-punning used all over the place).
552 This breaks on all non-gcc architectures (for now).
554 All those changes (plus the buffer of last time) allow me to gain 2 order
555 of magnitude on cruel tests consisting of 800000 array of integers on two
556 level of a hierarchical structure (200 secondes -> 4 secondes)
559 - the selector of reference must now return the type it points to, not
562 2004-08-06 Martin Quinson
563 Version 0.4 (protocol changed; ABI not changed)
564 - Allow to pass --gras-log argument to processes in simulation mode. Really.
565 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
566 - Implement a buffer transport, and use it by default (it relies on tcp in
567 real life and on sg in simulation).
568 That's a bit hackish since I had a new field to the structure to store
569 its data without interfering with the subtype ones. Inheritance
570 is tricky in C. And that's a kind of reverse inheritance with one class
571 derivating two classes. Or maybe a game with java interfaces. Anyway,
572 that's damn hard in C (at least).
573 Moreover, I got tired while trying to ensure plugin separation and
574 genericity in SG mode. MSG wants me to do weird things, so let's go for
575 cruel hacks (temporarily of course ;).
576 See comment in transport_private.h:71
577 - do not include all the _interface headers in private but in the files
578 which really need them (to cut the compilation time when they are
581 2004-07-26 Martin Quinson
582 Version 0.3 (protocol not changed; ABI changed)
583 - Major overhault of the datadesc interface to simplify it:
584 - shorted the function names:
585 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
586 - add a trivial way to push/pop integers into the cbps without malloc.
587 This allows to make really generic sub_type description, which simply
588 pop their size of the stack.
589 - add a function gras_datadesc_ref_pop_arr() which does what users want
590 most of the time: Declare a dynamic array (which pops its size of the
591 stack) and declare a reference to it. Poor name, but anyway.
592 - kill the post-send callback, add a post-receive one
594 2004-07-23 Martin Quinson
595 Version 0.2 (protocol changed; ABI changed)
596 - add some testing for cpbs in the test cases, and fix some more bugs.
597 This invalidate again the little64 data file, since I cannot regenerate
599 - remove an awfull optimization in the logging stuff, allowing me to:
600 - understand it again
601 - learn gcc how to check that the argument match the provided format
602 - fix all errors revealed by gcc after that
603 - internal keys of dict are not \0 terminated. Deal with it properly in
604 loggings instead of segfaulting when the user want to see the logs :-/
606 2004-07-22 Martin Quinson
607 - Fix some stupid bug preventing cbps (callback postit) from working
609 2004-07-21 Martin Quinson
610 - Some documentation cleanups
611 - remove the useless last argument of msgtype_declare
612 - rename the Virtu functions to fit into the 'os' namespace
613 - move headers src/include -> src/include/gras/ and stop fooling with
614 gras -> . symbolic link
615 - make distcheck is now successful
617 2004-07-19 Martin Quinson
619 - Build shared library also
620 - Install html doc to the right location
621 - stop removing maintainer files in make clean
622 - build tests only on make check
624 2004-07-13 Martin Quinson
626 - No major issue in previous version => change versionning schema
627 - Re-enable little64 convertion test now that Abdou kindly regenerated the
628 corresponding dataset.
630 2004-07-11 Martin Quinson
632 - Get it working with any kind of structure (we can compute the padding
633 bytes remotely for all the architectures I have access to)
634 - Implement the structure parsing macro (still not quite robust/complete)
635 - Improvement to the remote testing toysuite
637 2004-07-10 Martin Quinson
639 - get ride of a bunch of deprecated macros
640 - actually run the test for two-compliment, not only compile it :-/
641 - test whether the structures get packed (and bail out if yes. Damn.
642 Alignment is a serious matter)
643 - test whether the structures get compacted (but respecting the alignment
644 constraints of each types)
645 - test whether the array fields of structures can straddle alignment boundaries
647 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
648 (took me 2 hours to find that bug, looking at the wrong place)
649 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
650 before sending/receiving and are used to compute the offsets of fields
651 - Given that padding size depend even on compiler options, keep track of
652 alignment and aligned_size only for the current architecture. Not a big
653 deal since we send structure fields one after the other (seems
655 - Add the datastructure used for IEEE paper by the PBIO guys to the test
656 program, let it work on linux/gcc/little32. portability todo.
658 2004-07-08 Martin Quinson
659 - import and improve remote compilation support from FAST
660 - make sure make check works on half a dozen of machines out there
662 2004-07-07 Martin Quinson
663 Let's say it's version 0.0.3 ;)
664 - Implement conversions (yuhu!)
665 - Let it work on solaris (beside conversion, of course)
666 - Stupid me, using rand() to generate the conversion datatests in not wise.
668 2004-07-06 Martin Quinson
669 - Let make dist work, since I'm gonna need it to compile on remote hosts
670 - Let Tests/datadesc_usage write the architecture on which the file was
671 generated as first byte.
672 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
673 architecture descriptions.
674 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
675 on those architectures.
676 - Optimization: send/recv array of scalar in one shoot
678 2004-07-05 Martin Quinson
679 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
681 - Plug a whole bunch of memleaks
682 - each process now have to call gras_{init,exit}. One day, their log
683 settings will be separated
684 - Continue the code factorisation between SG, RL and common in Transport.
686 2004-07-04 Martin Quinson
688 - Redistribution between SG and RL.
689 We wanna have to accept in SG, so move accepted related parts of RL in
690 the common part. (more precisely, the dynar of all known sockets is no
691 more a static in transport.c, but part of the process_data)
693 [gras_stub_generator]
694 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
696 2004-07-03 Martin Quinson
697 - Create a new log channel tbx containing dict, set, log, dynar (to shut
698 them all up in one shot)
700 - Fix the ugly case of reference to dynamic array.
701 - New (semi-public) function gras_datadesc_size to allow the messaging
702 layer to malloc the needed space for the buffer.
704 - gras_socket_close now expect the socket to close (and not its address to
705 put NULL in it after it). This is because the socket passed to handlers
706 is one of their argument (=> not writable).
708 - propagate the interface cleanup from last week in datadesc, ie remove a
709 superfluous level of indirection. User pass adress of variable
710 containing data (both when sending and receiving), and not of a variable
711 being a pointer to the data. Let's say that I like it better ;)
712 The price for that is constructs like "int msg=*(int*)payload" in
713 handlers, but it's a fine price, IMHO.
715 - Let it work in RL (yuhu)
717 2004-06-21 Martin Quinson
719 - porting SG plugin and SG select to new standards (works almost).
720 - plug memleaks and fix bugs around.
723 - cleanup the prototype of data recv and force users to specify when they
724 want to handle references to objects. Test case working even for cycles.
725 - plug memleaks. Valgrind is perfectly ok with this.
727 2004-06-12 Martin Quinson
729 - cleanup the separation between plugin and main code in plugin creation
731 2004-06-11 Martin Quinson
733 - Reput hook for raw sockets, needed for BW experiments
734 - kill a few lines of dead code
735 [Data description] Interface cleanup
736 - gras_datadesc_by_name returns the searched type or NULL.
737 That way, no variable is needed to use a type desc once, which makes
739 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
740 two parameters were strings (field name, type name), leading to
742 [Dicos] Interface cleanup
743 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
744 This is consistant with the dynar API.
746 2004-04-21 Martin Quinson
748 - Porting to new standards.
751 There is no bag anymore, no need to take extra provision to mask the
752 pointers behind "ID".
753 Better splitup of functions between files create/exchange/convert.
754 This is still a bit artificial since convert and receive are so
755 interleaved, but anyway.
757 - add a queued message list to procdata (the ones not matching criteria
759 - factorize some more code between SG and RL wrt procdata
761 - use gras_exit in example to track memleaks
762 - get rid of gs_example now that GS is properly integrated into gras
763 - update run_test to integrate the lastest tests (datadesc)
765 - rename WARNINGn macros to WARNn since it prooved error-prone
767 2004-04-19 Martin Quinson
769 - register init/exit functions within gras module mechanism
770 - send/receive function.
771 Convertion is not implemented, but short-cutted if not needed.
772 struct/array elements are sent one by one (instead of block-wise), but
773 nobody really cares (yet). Get a prototype before optimizing.
774 - tests (using a file socket) for DD send/receive on:
775 - base types: int, float
776 - array: fixed size, string (ie ref to dynamic string)
777 - structure: homogeneous, heterogeneous
778 - chained list, graph with cycle
779 Believe it or not, valgrind is not too unhappy with the results. The
780 cycle happily segfaults, but the others are ok. And I'm sick of pointers
784 - Bugfix when using a filename explicitely (instead of '-')
786 2004-04-09 Martin Quinson
788 - factorize more code between RL and SG in socket creation
789 - Complete the implementation and tests of:
791 o file (only in RL, and mainly for debugging)
793 I lost 3 days to design a portable address resolver, and then decided
794 that the prototype mainly have to run on my box.
795 Addressing portability too early may be like optimizing too early :-/
797 - use gras_init in the Tests instead of the crappy parse_log_opt
798 (the latter function is removed)
799 [Conditional execution]
800 - New functions: gras_if_RL/gras_if_SG (basic support for this)
801 [Code reorganisation]
802 - Get rid of libgrasutils.a since it makes more trouble than it solves.
803 Build examples against the RL library, since there is no way to disable
804 its creation for now.