1 SimGrid (3.01) unstable; urgency=low
7 SimGrid (3.00) stable; urgency=low
10 * New! Give the possibility to hijack the surf parser and thus bypass
11 MSG_create_environment and MSG_launch_application. Have a look at
12 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
14 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
16 SimGrid (2.96) unstable; urgency=low
21 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
22 This deprecates the xbt_error_t mecanisms.
23 It modifies (simplifies) all XBT and GRAS API.
24 MSG API keeps unchanged (exceptions raised by XBT are catched from
25 within MSG and masked with existing error handling facilities)
28 * New! Add a FATPIPE model. [AL]
29 * New! Add a parallel task model. [AL]
30 * New! Add automatically a loopback interface (in the default
31 network model) if none was precised.
34 * Bugfix: MSG_process_resume now works with the current running process.
36 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
37 * Modification of MSG_task_get_compute_duration. Once a task has been
38 processed, the value returned by this function is now equal to 0. [AL]
39 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
40 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
41 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
42 of a m_task. For now, MSG_task_cancel only works with computation
44 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
45 of the processor (in Mflop/s) regardless of the current load on the
47 * API Change: use proper naming convention for MSG_getClock and
48 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
50 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
51 This function changes the priority of a computation task. This priority
52 doesn't affect the transfer rate. A priority of 2 will make a task
53 receive two times more cpu power than the other ones. This function
54 has been added to suit the needs of Nguyen The Loc and hasn't been that
55 much tested yet. So if it fails, please report it and send me your code.
57 * API Change: removed all functions and types that were marked "deprecated"
58 since many months. Renamed MSG_global_init_args to MSG_global_init.
60 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
62 SimGrid (2.95) unstable; urgency=low
65 * Steal some nice code to GNU pth to fix context detection and usage [AL]
66 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
67 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
70 * Bugfix: Allow absolute paths for platform description files [MQ]
71 * Bugfix: do free the variables after use. Leads to drastic performance
73 * Implement max_duration (ie, timeouts) on resources [AL]
76 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
78 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
79 the needed support (timeouts) [AL]
82 * Implement measurement sockets. You can now get the bandwidth between two
83 hosts thanks to AMOK (see below). [MQ]
84 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
85 dynar over the network (yeah) [MQ]
86 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
87 * simple/static token-ring example. [Alexandre Colucci and MQ]
88 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
92 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
93 type parsing. "short" and "long" modifiers were also ignored; other
94 modifier (such as reference level) are still ignored. [MQ]
95 * Embeed the buffer size within the buffer itself on SG. [MQ]
96 That way, send() are atomic and cannot get intermixed anymore (at least
97 the ones which are less than 100k; bigger messages still have the issue)
98 * Array size pushed by the field, not by the field type (or each
99 and every long int will push stuff to the cbps) [MQ]
100 * use select() to sleep since it allows to portably sleep less than one
103 GRAS (minor cleanups)
104 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
107 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
108 * type category 'ignored' killed as it was never used and were difficult
110 * whether a type can cycle or not is now a flag, leaving room for more
111 flags (as "ignored", if we feel the need one day ;)
112 * Rename raw sockets to measurement sockets since "raw" has another
113 meaning in networking community.
116 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
117 and offers features not belonging to GRAS but that most applications
118 need. One day, it may be a set of plugins loadable at runtime.
119 * New module: bandwidth
120 bandwidth measurement between arbitrary nodes running this module. [MQ]
122 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
124 SimGrid (2.94) unstable; urgency=low
126 The first beta release of SimGrid 3 !
130 * Update the main page and the FAQ. Adding references to gforge.
133 * Add a gras_os_getpid function.
136 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
137 * Extend the logs so that they also print PID, hostname, date, ... if
139 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
140 and kill the old version which were in testsuite/
141 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
145 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
146 for users. Instead, each programming environment (GRAS, MSG,...) use it
147 when needed to provide such a feature to users.
148 Don't shortcut the mecanism or you will also shortcut the virtualization
149 you need on the simulator.
153 * Cleanups in configury with regard to compile optimization/warning flags.
154 Also add -fno-loop-optimize to any powerpc since it's the optimization
155 killing gcc (< 3.4.0).
156 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
157 needed by MSG examples complications
158 * Borrow configury beautifications from PHP
161 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
162 warning (thanks loris for stumbling into this one).
163 * Bugfix: stop loading private headers (gras_config.h) from the public
167 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
168 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
169 RL would imply a DNS resolver, which is *hard* to do in a portable way
170 (and therefore delayed).
171 * Implement a real timer mecanism and use it in timing macros. This allows
172 to avoid rounding errors and get a 0.000005 sec precision in timing
173 macros. While I was at it, various cleanups:
174 - allow to declare more than one timed section per file (fix a stupid bug)
175 - move some private declaration to the right place
176 - merge conditional execution and timing macros into emulation module
177 - document the module
178 - make sure the module cleanups its mess on gras_exit
179 * Documentation improvements:
180 - (new) how to compile applications using GRAS
181 - (new) emulation support (timing macros)
183 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
185 SimGrid (2.93) unstable; urgency=low
187 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
190 - Use Paje properly where used. Still to be sanitized properly.
191 - Portability fix: Add an implementation of the contexts using pthread
195 - Add xbt_procname(): returns the name of the current process.
196 Use it to show the current process's name in all logging.
198 - fix detection of older flex version and the reaction, since we do
199 depend on modern ones (we use lex_destroy)
200 - Better separation of SG and RL in the libs: remove all simulation code
201 from libgras. As a result, this lib is now only 200k when stripped.
202 Some of the xbt modules may also be duplicated (two sets and such) and
203 should be cleaned/killed before SG3.
204 - Insist on using xlC on AIX because of weird problems involving gcc there.
205 - Cleanup the make remote stuff. This is now done by scripts
206 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
207 mainly for our private use, but we're working on changing them to user
210 - Bugfix: flush the socket on close only if there is some *output*.
211 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
213 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
216 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
218 SimGrid (2.92) unstable; urgency=low
220 Alpha 3 on the path to SimGrid 3
224 - New! First try of benchmarking macros.
225 - New! First try so that gras_stub_generator generate deployment and
226 remote compilation helpers.
228 - Bugfix: Initialization fix in msg_test.
232 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
234 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
236 - API Change: gras_os_sleep to take the amount of seconds as a double.
237 Accepting an int was error prone since it was the only location where
238 seconds were coded as such. It leaded to damn rounding errors.
239 - Bugfix: Hard to belive that timers ever worked before this.
241 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
243 SimGrid (2.91) unstable; urgency=low
245 Alpha 2 on the path to SimGrid 3
249 - Bug fix in the lmm_solver.
251 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
252 through the function MSG_paje_output.
253 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
254 - It is possible to bound the rate of a communication in MSG with
255 MSG_task_put_bounded() (was already in the previous version but I had forgotten
256 to write it in the changelog).
257 - Bug fix to let GRAS run on top of MSG until we move it directly on top
262 - Various cleanups to the autotools stuff
263 - Begin to move Gras examples to examples/gras/
264 - Let make distcheck work again (yeah!)
266 - documentation overhauled using doxygen.
267 gtk-doc-tools is dead in SimGrid now.
268 - Automatically extract all existing logging categories, and add the list
269 to the documentation (long standing one, to say the less)
271 - Cleanup the known architecture table. Reorder the entries to group what
272 should be, and use a more consistent naming scheme.
273 (some of the test dataset are still to be regenerated)
274 - New! Allow library to register globals on each process just as userdata
276 This is implemented using a xbt_dict and not a xbt_set, so we loose the
277 lookup time (for now).
278 Use it in msg and trp.
279 This cleans a lot the internals and helps enforcing privacy of the
280 headers between the gras components.
281 - New! Add a timer mechanism, not unlike cron(8) and at(1).
282 - Bugfix: gras_os_time was delirious in RL.
283 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
285 - Reenable GRAS now that it works.
287 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
289 SimGrid (2.90) unstable; urgency=low
291 Alpha 1 on the path to SimGrid 3
293 * It is a long time since the last release of SimGrid. I'm sorry about
294 that but as I had told you, I was rewriting a lot of things. I apologize
295 to those who had been reporting bugs to me and that I had not answered.
296 If your bug is still in the new version, please tell me. Here is a
297 summary of the main changes.
299 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
300 lead by Martin Quinson. As a consequence SimGrid gains a lot in
301 portability, speed, and a lot more but you'll figure it out later.
302 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
303 I wanted to release the new MSG as soon as possible and I have
304 broken GRAS, which is the reason why, for now, only MSG is fully
305 functional. A laconic description of these projects is available
306 in the documentation.
308 * REVOLUTION 2: I have removed SG and I am now using a new simulation
309 kernel optimized for our needs (called SURF but only the developers
310 should use it). Hence, MSG is now roughly 30 times faster and I think
311 that by rewriting a little bit MSG, I could event speed it up a little
312 bit more. Beside the gain in speed, it is also much easier to encode a
313 new platform model with SURF than it was with SG. More to come...
315 * REVOLUTION 3: I have tried to change a little as possible the API of
316 MSG but a few things really had to disappear. The main differences
317 with the previous version are :
318 1) no more m_links_t and the corresponding functions. Platforms are
319 directly read from a XML description and cannot be hard-coded
320 anymore. The same format is used for application deployment
321 description. The new format is described in the documentation.
322 Have a look in tools/platform_generation. There is a tiny script
323 that converts from the old platform format to the new one. Concerning
324 the application deployment format, parsing the old one is tricky.
325 I think most of you should however be able to convert your files. If
326 it is really an issue, I can write a C code that does the conversion.
328 2) the toolbox tbx does not exist anymore. We now have a library
329 with much more data-structures but without the hash-tables (we have
330 dictionaries that are much faster).
332 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
334 *****************************************************************************
335 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
336 * don't feel like converting the oldies. (Mt) *
337 *****************************************************************************
340 Version 2.90: "the long awaited one"
341 - Finished rewriting and debugging MSG. Rewrote the documentation.
342 - disable GRAS for now since it needs to be ported to the newest SG
345 - Finish the port to windows (using mingw32 for cross-compile)
348 - Main loop and datastructures of SURF. A cpu resource object is
349 functional. Surf can thus be used to create cpu's with variable
350 performance on which you can execute some actions.
352 2004-11-15 Martin Quinson
353 - Port to ARM. Simply added the alignment and size descriptions. Should
354 work, but the ARM machines are so slow that I didn't had the opportunity
355 to 'make check' over there yet.
357 2004-11-15 Arnaud Legrand
358 - Trace manager now written. It uses a heap structure and is therefore
359 expected to be efficient. It may however be speeded up (particularly
360 when many events occur at the same date) by using red and black
361 trees. One day maybe...
362 - Max-min linear system solver written. It uses a sparse matrix
363 structure taking advantage of its expected use. Most operations are
364 O(1) and free/calloc are called as few as possible. The computation of
365 the minimum could however be improved by using a red and black tree
368 2004-11-03 Arnaud Legrand
369 - Rename every gras_* function that was in xbt/ to its xbt_
371 - Add a heap and a doubly-linked list to xbt
372 - Added a dichotomy to the dictionaries. make check works as well before
373 so I assume that the patch is correct. I do not know however if things
374 run effectively faster than before now. :)
376 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
377 SimGrid, and the version number is bumped to 2.x
379 2004-10-29 Martin Quinson
380 - Introduction of the remote errors.
381 They are the result of a RMI/RPC on the remote machine.
382 ErrCodes being scalar values, you can't get the host on which those
383 errors did happen. Extending the error mechanism as in Gnome is possible.
384 No idea yet whether it is a good idea.
386 2004-10-28 Martin Quinson
387 - Interface revolution: the Starred Structure Eradication.
388 I used to do typedef struct {} toto_t; and then handle *toto_t.
389 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
390 - ???_t is a valid type (builded with typedef)
391 - s_toto_t is a structure (access to fields with .)
392 - s_toto is a structure needing 'struct' keyword to be used
393 - e_toto_t is an enum
394 - toto_t is an 'object' (struct*)
396 typedef struct s_toto {} s_toto_t, *toto_t;
397 typedef enum {} e_toto_t;
398 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
399 s_toto_t) is private.
401 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
402 so that it can set the variable to NULL. It was so for dicts and sets,
403 it changed for dynars.
405 - Fix a bunch of memleaks in dict_remove
406 - Fix a bug in sg/server_socket opening: it failed all the time.
408 2004-10-07 Martin Quinson
409 - Speed up dynar lookup operation a bit.
411 gras_dynar_get is dead.
413 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
414 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
415 (which gives you the address of the stored data).
417 gras_dynar_get_as is an helpful macro which allows you to retrieve a
418 copy of the data using an affectation to do the job and not a memcpy.
420 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
421 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
423 It does not really speedup the dynar test because they are
424 setting elements all the time (and look them seldom). But the dict does
425 far more lookup than setting.
427 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
429 2004-10-05 Martin Quinson
430 - Allow to (en/dis)able the cycle detection at run time.
432 Whether we should check for cycle or not is now a property of each
433 datatype. When you think there may be some cycle, use datadesc_cycle_set.
434 datadesc_cycle_unset allow to remove this property when previously set.
436 Note that the cycle detection is off by default since it impacts the
437 performance. Watch the data you feed GRAS with ;)
439 This property is hereditary. Any element embedded in a structure having it
440 set have it set for the time of this data exchange.
442 You should set it both on sender and receiver side. If you don't set it on
443 sender side, it will enter an endless loop. If you forget on receiver
444 side, the cycles won't be recreated after communication.
446 - Header reorganization.
447 Kill gras_private.h, each submodule must load the headers it needs.
449 2004-10-04 Martin Quinson
450 - Interface revolution: do not try to survive to malloc failure.
452 Now, gras_malloc and friends call gras_abort() on failure.
453 As a conclusion, malloc_error is not a valid error anymore, and all
454 functions for which it was the only gras_error_t return value are
455 changed. They now return void, or there result directly.
456 This simplify the API a lot.
458 2004-09-29 Martin Quinson
459 - Re-enable raw sockets.
460 Created by gras_socket_{client,server}_ext;
461 Used with gras_raw_{send,recv}
464 It should allow to kill the last bits of gras first version soon.
466 This is not completely satisfactory yet (duplicate code with
467 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
470 - Simplify transport plugin (internal) interface by not passing any
471 argument to _server and _client, but embedding them in the socket
474 2004-09-28 Martin Quinson
475 - Finish the port to AIX.
476 autoconf was my problem (segfault within the malloc replacement
477 function. No idea why)
479 2004-09-16 Martin Quinson
480 - Fix some size_t madness on 64bit architectures.
482 2004-09-08 Martin Quinson
483 - Reduce the number of system headers loaded, overload some more system
484 calls (such as malloc to cast the result of the system one, and work
486 - Fix and reintroduce the config support
488 2004-09-07 Martin Quinson
489 - Source code reorganization to allow Arnaud to surf all over there.
490 - Allow to document the logging categories.
491 - Remove all uppercase from logging categories and useless cleanup in names.
493 2004-08-18 Martin Quinson
494 Version 0.6.2 (protocol not changed; API changed)
495 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
496 gras_error_t), and NULL when not found. Functions expecting a msgtype
497 as argument (msg_wait; msg_send) deal with NULL argument by providing a
498 hopefully usefull message.
499 - Portability to prehistoric sparcs again
501 2004-08-17 Martin Quinson
502 Version 0.6.1 (protocol not changed; ABI not changed)
503 - prealloc some buffers to speed things up
505 2004-08-11 Martin Quinson
506 Version 0.6 (protocol not changed; ABI expended)
507 - The parsing macro can deal with the references, provided that you add
508 the relevant annotations (using GRAS_ANNOTE(size,field_name))
510 2004-08-09 Martin Quinson
511 Version 0.5 (protocol not changed; ABI changed)
512 - Allow to off turn the cycle detection code in data exchange at
513 compilation time. It should be at run time, but I'm short of time (and
514 the config stuff is still broken). That way, we keep dict out of the
515 critical path, which is good because the performance is poor:
516 - search not dichotomial yet
517 - dynar give no way to access their content and memcpy everytime
518 - In composed data description (struct, ref and so on), stop foolness of
519 keeping the subtype's ID, but store the type itself. This keeps sets out
520 of the critical path, which is good since they rely on dynar and
521 dictionnaries. The only loose of that is that we cannot detect the
522 redeclaration of a structure/union with another content (but I'm not sure
523 the code detected well this error before anyway). We still can detect
524 the redefinition discrepancy for the other types.
525 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
526 it breaks the code because of type-punning used all over the place).
527 This breaks on all non-gcc architectures (for now).
529 All those changes (plus the buffer of last time) allow me to gain 2 order
530 of magnitude on cruel tests consisting of 800000 array of integers on two
531 level of a hierarchical structure (200 secondes -> 4 secondes)
534 - the selector of reference must now return the type it points to, not
537 2004-08-06 Martin Quinson
538 Version 0.4 (protocol changed; ABI not changed)
539 - Allow to pass --gras-log argument to processes in simulation mode. Really.
540 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
541 - Implement a buffer transport, and use it by default (it relies on tcp in
542 real life and on sg in simulation).
543 That's a bit hackish since I had a new field to the structure to store
544 its data without interfering with the subtype ones. Inheritance
545 is tricky in C. And that's a kind of reverse inheritance with one class
546 derivating two classes. Or maybe a game with java interfaces. Anyway,
547 that's damn hard in C (at least).
548 Moreover, I got tired while trying to ensure plugin separation and
549 genericity in SG mode. MSG wants me to do weird things, so let's go for
550 cruel hacks (temporarily of course ;).
551 See comment in transport_private.h:71
552 - do not include all the _interface headers in private but in the files
553 which really need them (to cut the compilation time when they are
556 2004-07-26 Martin Quinson
557 Version 0.3 (protocol not changed; ABI changed)
558 - Major overhault of the datadesc interface to simplify it:
559 - shorted the function names:
560 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
561 - add a trivial way to push/pop integers into the cbps without malloc.
562 This allows to make really generic sub_type description, which simply
563 pop their size of the stack.
564 - add a function gras_datadesc_ref_pop_arr() which does what users want
565 most of the time: Declare a dynamic array (which pops its size of the
566 stack) and declare a reference to it. Poor name, but anyway.
567 - kill the post-send callback, add a post-receive one
569 2004-07-23 Martin Quinson
570 Version 0.2 (protocol changed; ABI changed)
571 - add some testing for cpbs in the test cases, and fix some more bugs.
572 This invalidate again the little64 data file, since I cannot regenerate
574 - remove an awfull optimization in the logging stuff, allowing me to:
575 - understand it again
576 - learn gcc how to check that the argument match the provided format
577 - fix all errors revealed by gcc after that
578 - internal keys of dict are not \0 terminated. Deal with it properly in
579 loggings instead of segfaulting when the user want to see the logs :-/
581 2004-07-22 Martin Quinson
582 - Fix some stupid bug preventing cbps (callback postit) from working
584 2004-07-21 Martin Quinson
585 - Some documentation cleanups
586 - remove the useless last argument of msgtype_declare
587 - rename the Virtu functions to fit into the 'os' namespace
588 - move headers src/include -> src/include/gras/ and stop fooling with
589 gras -> . symbolic link
590 - make distcheck is now successful
592 2004-07-19 Martin Quinson
594 - Build shared library also
595 - Install html doc to the right location
596 - stop removing maintainer files in make clean
597 - build tests only on make check
599 2004-07-13 Martin Quinson
601 - No major issue in previous version => change versionning schema
602 - Re-enable little64 convertion test now that Abdou kindly regenerated the
603 corresponding dataset.
605 2004-07-11 Martin Quinson
607 - Get it working with any kind of structure (we can compute the padding
608 bytes remotely for all the architectures I have access to)
609 - Implement the structure parsing macro (still not quite robust/complete)
610 - Improvement to the remote testing toysuite
612 2004-07-10 Martin Quinson
614 - get ride of a bunch of deprecated macros
615 - actually run the test for two-compliment, not only compile it :-/
616 - test whether the structures get packed (and bail out if yes. Damn.
617 Alignment is a serious matter)
618 - test whether the structures get compacted (but respecting the alignment
619 constraints of each types)
620 - test whether the array fields of structures can straddle alignment boundaries
622 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
623 (took me 2 hours to find that bug, looking at the wrong place)
624 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
625 before sending/receiving and are used to compute the offsets of fields
626 - Given that padding size depend even on compiler options, keep track of
627 alignment and aligned_size only for the current architecture. Not a big
628 deal since we send structure fields one after the other (seems
630 - Add the datastructure used for IEEE paper by the PBIO guys to the test
631 program, let it work on linux/gcc/little32. portability todo.
633 2004-07-08 Martin Quinson
634 - import and improve remote compilation support from FAST
635 - make sure make check works on half a dozen of machines out there
637 2004-07-07 Martin Quinson
638 Let's say it's version 0.0.3 ;)
639 - Implement conversions (yuhu!)
640 - Let it work on solaris (beside conversion, of course)
641 - Stupid me, using rand() to generate the conversion datatests in not wise.
643 2004-07-06 Martin Quinson
644 - Let make dist work, since I'm gonna need it to compile on remote hosts
645 - Let Tests/datadesc_usage write the architecture on which the file was
646 generated as first byte.
647 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
648 architecture descriptions.
649 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
650 on those architectures.
651 - Optimization: send/recv array of scalar in one shoot
653 2004-07-05 Martin Quinson
654 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
656 - Plug a whole bunch of memleaks
657 - each process now have to call gras_{init,exit}. One day, their log
658 settings will be separated
659 - Continue the code factorisation between SG, RL and common in Transport.
661 2004-07-04 Martin Quinson
663 - Redistribution between SG and RL.
664 We wanna have to accept in SG, so move accepted related parts of RL in
665 the common part. (more precisely, the dynar of all known sockets is no
666 more a static in transport.c, but part of the process_data)
668 [gras_stub_generator]
669 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
671 2004-07-03 Martin Quinson
672 - Create a new log channel tbx containing dict, set, log, dynar (to shut
673 them all up in one shot)
675 - Fix the ugly case of reference to dynamic array.
676 - New (semi-public) function gras_datadesc_size to allow the messaging
677 layer to malloc the needed space for the buffer.
679 - gras_socket_close now expect the socket to close (and not its address to
680 put NULL in it after it). This is because the socket passed to handlers
681 is one of their argument (=> not writable).
683 - propagate the interface cleanup from last week in datadesc, ie remove a
684 superfluous level of indirection. User pass adress of variable
685 containing data (both when sending and receiving), and not of a variable
686 being a pointer to the data. Let's say that I like it better ;)
687 The price for that is constructs like "int msg=*(int*)payload" in
688 handlers, but it's a fine price, IMHO.
690 - Let it work in RL (yuhu)
692 2004-06-21 Martin Quinson
694 - porting SG plugin and SG select to new standards (works almost).
695 - plug memleaks and fix bugs around.
698 - cleanup the prototype of data recv and force users to specify when they
699 want to handle references to objects. Test case working even for cycles.
700 - plug memleaks. Valgrind is perfectly ok with this.
702 2004-06-12 Martin Quinson
704 - cleanup the separation between plugin and main code in plugin creation
706 2004-06-11 Martin Quinson
708 - Reput hook for raw sockets, needed for BW experiments
709 - kill a few lines of dead code
710 [Data description] Interface cleanup
711 - gras_datadesc_by_name returns the searched type or NULL.
712 That way, no variable is needed to use a type desc once, which makes
714 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
715 two parameters were strings (field name, type name), leading to
717 [Dicos] Interface cleanup
718 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
719 This is consistant with the dynar API.
721 2004-04-21 Martin Quinson
723 - Porting to new standards.
726 There is no bag anymore, no need to take extra provision to mask the
727 pointers behind "ID".
728 Better splitup of functions between files create/exchange/convert.
729 This is still a bit artificial since convert and receive are so
730 interleaved, but anyway.
732 - add a queued message list to procdata (the ones not matching criteria
734 - factorize some more code between SG and RL wrt procdata
736 - use gras_exit in example to track memleaks
737 - get rid of gs_example now that GS is properly integrated into gras
738 - update run_test to integrate the lastest tests (datadesc)
740 - rename WARNINGn macros to WARNn since it prooved error-prone
742 2004-04-19 Martin Quinson
744 - register init/exit functions within gras module mechanism
745 - send/receive function.
746 Convertion is not implemented, but short-cutted if not needed.
747 struct/array elements are sent one by one (instead of block-wise), but
748 nobody really cares (yet). Get a prototype before optimizing.
749 - tests (using a file socket) for DD send/receive on:
750 - base types: int, float
751 - array: fixed size, string (ie ref to dynamic string)
752 - structure: homogeneous, heterogeneous
753 - chained list, graph with cycle
754 Believe it or not, valgrind is not too unhappy with the results. The
755 cycle happily segfaults, but the others are ok. And I'm sick of pointers
759 - Bugfix when using a filename explicitely (instead of '-')
761 2004-04-09 Martin Quinson
763 - factorize more code between RL and SG in socket creation
764 - Complete the implementation and tests of:
766 o file (only in RL, and mainly for debugging)
768 I lost 3 days to design a portable address resolver, and then decided
769 that the prototype mainly have to run on my box.
770 Addressing portability too early may be like optimizing too early :-/
772 - use gras_init in the Tests instead of the crappy parse_log_opt
773 (the latter function is removed)
774 [Conditional execution]
775 - New functions: gras_if_RL/gras_if_SG (basic support for this)
776 [Code reorganisation]
777 - Get rid of libgrasutils.a since it makes more trouble than it solves.
778 Build examples against the RL library, since there is no way to disable
779 its creation for now.