1 SimGrid (2.97) unstable; urgency=low
4 * New! Give the possibility to hijack the surf parser and thus bypass
5 MSG_create_environment and MSG_launch_application. Have a look at
6 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
10 SimGrid (2.96) unstable; urgency=low
15 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
16 This deprecates the xbt_error_t mecanisms.
17 It modifies (simplifies) all XBT and GRAS API.
18 MSG API keeps unchanged (exceptions raised by XBT are catched from
19 within MSG and masked with existing error handling facilities)
22 * New! Add a FATPIPE model. [AL]
23 * New! Add a parallel task model. [AL]
24 * New! Add automatically a loopback interface (in the default
25 network model) if none was precised.
28 * Bugfix: MSG_process_resume now works with the current running process.
30 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
31 * Modification of MSG_task_get_compute_duration. Once a task has been
32 processed, the value returned by this function is now equal to 0. [AL]
33 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
34 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
35 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
36 of a m_task. For now, MSG_task_cancel only works with computation
38 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
39 of the processor (in Mflop/s) regardless of the current load on the
41 * API Change: use proper naming convention for MSG_getClock and
42 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
44 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
45 This function changes the priority of a computation task. This priority
46 doesn't affect the transfer rate. A priority of 2 will make a task
47 receive two times more cpu power than the other ones. This function
48 has been added to suit the needs of Nguyen The Loc and hasn't been that
49 much tested yet. So if it fails, please report it and send me your code.
51 * API Change: removed all functions and types that were marked "deprecated"
52 since many months. Renamed MSG_global_init_args to MSG_global_init.
54 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
56 SimGrid (2.95) unstable; urgency=low
59 * Steal some nice code to GNU pth to fix context detection and usage [AL]
60 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
61 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
64 * Bugfix: Allow absolute paths for platform description files [MQ]
65 * Bugfix: do free the variables after use. Leads to drastic performance
67 * Implement max_duration (ie, timeouts) on resources [AL]
70 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
72 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
73 the needed support (timeouts) [AL]
76 * Implement measurement sockets. You can now get the bandwidth between two
77 hosts thanks to AMOK (see below). [MQ]
78 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
79 dynar over the network (yeah) [MQ]
80 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
81 * simple/static token-ring example. [Alexandre Colucci and MQ]
82 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
86 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
87 type parsing. "short" and "long" modifiers were also ignored; other
88 modifier (such as reference level) are still ignored. [MQ]
89 * Embeed the buffer size within the buffer itself on SG. [MQ]
90 That way, send() are atomic and cannot get intermixed anymore (at least
91 the ones which are less than 100k; bigger messages still have the issue)
92 * Array size pushed by the field, not by the field type (or each
93 and every long int will push stuff to the cbps) [MQ]
94 * use select() to sleep since it allows to portably sleep less than one
98 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
101 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
102 * type category 'ignored' killed as it was never used and were difficult
104 * whether a type can cycle or not is now a flag, leaving room for more
105 flags (as "ignored", if we feel the need one day ;)
106 * Rename raw sockets to measurement sockets since "raw" has another
107 meaning in networking community.
110 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
111 and offers features not belonging to GRAS but that most applications
112 need. One day, it may be a set of plugins loadable at runtime.
113 * New module: bandwidth
114 bandwidth measurement between arbitrary nodes running this module. [MQ]
116 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
118 SimGrid (2.94) unstable; urgency=low
120 The first beta release of SimGrid 3 !
124 * Update the main page and the FAQ. Adding references to gforge.
127 * Add a gras_os_getpid function.
130 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
131 * Extend the logs so that they also print PID, hostname, date, ... if
133 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
134 and kill the old version which were in testsuite/
135 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
139 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
140 for users. Instead, each programming environment (GRAS, MSG,...) use it
141 when needed to provide such a feature to users.
142 Don't shortcut the mecanism or you will also shortcut the virtualization
143 you need on the simulator.
147 * Cleanups in configury with regard to compile optimization/warning flags.
148 Also add -fno-loop-optimize to any powerpc since it's the optimization
149 killing gcc (< 3.4.0).
150 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
151 needed by MSG examples complications
152 * Borrow configury beautifications from PHP
155 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
156 warning (thanks loris for stumbling into this one).
157 * Bugfix: stop loading private headers (gras_config.h) from the public
161 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
162 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
163 RL would imply a DNS resolver, which is *hard* to do in a portable way
164 (and therefore delayed).
165 * Implement a real timer mecanism and use it in timing macros. This allows
166 to avoid rounding errors and get a 0.000005 sec precision in timing
167 macros. While I was at it, various cleanups:
168 - allow to declare more than one timed section per file (fix a stupid bug)
169 - move some private declaration to the right place
170 - merge conditional execution and timing macros into emulation module
171 - document the module
172 - make sure the module cleanups its mess on gras_exit
173 * Documentation improvements:
174 - (new) how to compile applications using GRAS
175 - (new) emulation support (timing macros)
177 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
179 SimGrid (2.93) unstable; urgency=low
181 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
184 - Use Paje properly where used. Still to be sanitized properly.
185 - Portability fix: Add an implementation of the contexts using pthread
189 - Add xbt_procname(): returns the name of the current process.
190 Use it to show the current process's name in all logging.
192 - fix detection of older flex version and the reaction, since we do
193 depend on modern ones (we use lex_destroy)
194 - Better separation of SG and RL in the libs: remove all simulation code
195 from libgras. As a result, this lib is now only 200k when stripped.
196 Some of the xbt modules may also be duplicated (two sets and such) and
197 should be cleaned/killed before SG3.
198 - Insist on using xlC on AIX because of weird problems involving gcc there.
199 - Cleanup the make remote stuff. This is now done by scripts
200 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
201 mainly for our private use, but we're working on changing them to user
204 - Bugfix: flush the socket on close only if there is some *output*.
205 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
207 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
210 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
212 SimGrid (2.92) unstable; urgency=low
214 Alpha 3 on the path to SimGrid 3
218 - New! First try of benchmarking macros.
219 - New! First try so that gras_stub_generator generate deployment and
220 remote compilation helpers.
222 - Bugfix: Initialization fix in msg_test.
226 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
228 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
230 - API Change: gras_os_sleep to take the amount of seconds as a double.
231 Accepting an int was error prone since it was the only location where
232 seconds were coded as such. It leaded to damn rounding errors.
233 - Bugfix: Hard to belive that timers ever worked before this.
235 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
237 SimGrid (2.91) unstable; urgency=low
239 Alpha 2 on the path to SimGrid 3
243 - Bug fix in the lmm_solver.
245 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
246 through the function MSG_paje_output.
247 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
248 - It is possible to bound the rate of a communication in MSG with
249 MSG_task_put_bounded() (was already in the previous version but I had forgotten
250 to write it in the changelog).
251 - Bug fix to let GRAS run on top of MSG until we move it directly on top
256 - Various cleanups to the autotools stuff
257 - Begin to move Gras examples to examples/gras/
258 - Let make distcheck work again (yeah!)
260 - documentation overhauled using doxygen.
261 gtk-doc-tools is dead in SimGrid now.
262 - Automatically extract all existing logging categories, and add the list
263 to the documentation (long standing one, to say the less)
265 - Cleanup the known architecture table. Reorder the entries to group what
266 should be, and use a more consistent naming scheme.
267 (some of the test dataset are still to be regenerated)
268 - New! Allow library to register globals on each process just as userdata
270 This is implemented using a xbt_dict and not a xbt_set, so we loose the
271 lookup time (for now).
272 Use it in msg and trp.
273 This cleans a lot the internals and helps enforcing privacy of the
274 headers between the gras components.
275 - New! Add a timer mechanism, not unlike cron(8) and at(1).
276 - Bugfix: gras_os_time was delirious in RL.
277 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
279 - Reenable GRAS now that it works.
281 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
283 SimGrid (2.90) unstable; urgency=low
285 Alpha 1 on the path to SimGrid 3
287 * It is a long time since the last release of SimGrid. I'm sorry about
288 that but as I had told you, I was rewriting a lot of things. I apologize
289 to those who had been reporting bugs to me and that I had not answered.
290 If your bug is still in the new version, please tell me. Here is a
291 summary of the main changes.
293 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
294 lead by Martin Quinson. As a consequence SimGrid gains a lot in
295 portability, speed, and a lot more but you'll figure it out later.
296 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
297 I wanted to release the new MSG as soon as possible and I have
298 broken GRAS, which is the reason why, for now, only MSG is fully
299 functional. A laconic description of these projects is available
300 in the documentation.
302 * REVOLUTION 2: I have removed SG and I am now using a new simulation
303 kernel optimized for our needs (called SURF but only the developers
304 should use it). Hence, MSG is now roughly 30 times faster and I think
305 that by rewriting a little bit MSG, I could event speed it up a little
306 bit more. Beside the gain in speed, it is also much easier to encode a
307 new platform model with SURF than it was with SG. More to come...
309 * REVOLUTION 3: I have tried to change a little as possible the API of
310 MSG but a few things really had to disappear. The main differences
311 with the previous version are :
312 1) no more m_links_t and the corresponding functions. Platforms are
313 directly read from a XML description and cannot be hard-coded
314 anymore. The same format is used for application deployment
315 description. The new format is described in the documentation.
316 Have a look in tools/platform_generation. There is a tiny script
317 that converts from the old platform format to the new one. Concerning
318 the application deployment format, parsing the old one is tricky.
319 I think most of you should however be able to convert your files. If
320 it is really an issue, I can write a C code that does the conversion.
322 2) the toolbox tbx does not exist anymore. We now have a library
323 with much more data-structures but without the hash-tables (we have
324 dictionaries that are much faster).
326 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
328 *****************************************************************************
329 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
330 * don't feel like converting the oldies. (Mt) *
331 *****************************************************************************
334 Version 2.90: "the long awaited one"
335 - Finished rewriting and debugging MSG. Rewrote the documentation.
336 - disable GRAS for now since it needs to be ported to the newest SG
339 - Finish the port to windows (using mingw32 for cross-compile)
342 - Main loop and datastructures of SURF. A cpu resource object is
343 functional. Surf can thus be used to create cpu's with variable
344 performance on which you can execute some actions.
346 2004-11-15 Martin Quinson
347 - Port to ARM. Simply added the alignment and size descriptions. Should
348 work, but the ARM machines are so slow that I didn't had the opportunity
349 to 'make check' over there yet.
351 2004-11-15 Arnaud Legrand
352 - Trace manager now written. It uses a heap structure and is therefore
353 expected to be efficient. It may however be speeded up (particularly
354 when many events occur at the same date) by using red and black
355 trees. One day maybe...
356 - Max-min linear system solver written. It uses a sparse matrix
357 structure taking advantage of its expected use. Most operations are
358 O(1) and free/calloc are called as few as possible. The computation of
359 the minimum could however be improved by using a red and black tree
362 2004-11-03 Arnaud Legrand
363 - Rename every gras_* function that was in xbt/ to its xbt_
365 - Add a heap and a doubly-linked list to xbt
366 - Added a dichotomy to the dictionaries. make check works as well before
367 so I assume that the patch is correct. I do not know however if things
368 run effectively faster than before now. :)
370 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
371 SimGrid, and the version number is bumped to 2.x
373 2004-10-29 Martin Quinson
374 - Introduction of the remote errors.
375 They are the result of a RMI/RPC on the remote machine.
376 ErrCodes being scalar values, you can't get the host on which those
377 errors did happen. Extending the error mechanism as in Gnome is possible.
378 No idea yet whether it is a good idea.
380 2004-10-28 Martin Quinson
381 - Interface revolution: the Starred Structure Eradication.
382 I used to do typedef struct {} toto_t; and then handle *toto_t.
383 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
384 - ???_t is a valid type (builded with typedef)
385 - s_toto_t is a structure (access to fields with .)
386 - s_toto is a structure needing 'struct' keyword to be used
387 - e_toto_t is an enum
388 - toto_t is an 'object' (struct*)
390 typedef struct s_toto {} s_toto_t, *toto_t;
391 typedef enum {} e_toto_t;
392 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
393 s_toto_t) is private.
395 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
396 so that it can set the variable to NULL. It was so for dicts and sets,
397 it changed for dynars.
399 - Fix a bunch of memleaks in dict_remove
400 - Fix a bug in sg/server_socket opening: it failed all the time.
402 2004-10-07 Martin Quinson
403 - Speed up dynar lookup operation a bit.
405 gras_dynar_get is dead.
407 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
408 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
409 (which gives you the address of the stored data).
411 gras_dynar_get_as is an helpful macro which allows you to retrieve a
412 copy of the data using an affectation to do the job and not a memcpy.
414 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
415 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
417 It does not really speedup the dynar test because they are
418 setting elements all the time (and look them seldom). But the dict does
419 far more lookup than setting.
421 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
423 2004-10-05 Martin Quinson
424 - Allow to (en/dis)able the cycle detection at run time.
426 Whether we should check for cycle or not is now a property of each
427 datatype. When you think there may be some cycle, use datadesc_cycle_set.
428 datadesc_cycle_unset allow to remove this property when previously set.
430 Note that the cycle detection is off by default since it impacts the
431 performance. Watch the data you feed GRAS with ;)
433 This property is hereditary. Any element embedded in a structure having it
434 set have it set for the time of this data exchange.
436 You should set it both on sender and receiver side. If you don't set it on
437 sender side, it will enter an endless loop. If you forget on receiver
438 side, the cycles won't be recreated after communication.
440 - Header reorganization.
441 Kill gras_private.h, each submodule must load the headers it needs.
443 2004-10-04 Martin Quinson
444 - Interface revolution: do not try to survive to malloc failure.
446 Now, gras_malloc and friends call gras_abort() on failure.
447 As a conclusion, malloc_error is not a valid error anymore, and all
448 functions for which it was the only gras_error_t return value are
449 changed. They now return void, or there result directly.
450 This simplify the API a lot.
452 2004-09-29 Martin Quinson
453 - Re-enable raw sockets.
454 Created by gras_socket_{client,server}_ext;
455 Used with gras_raw_{send,recv}
458 It should allow to kill the last bits of gras first version soon.
460 This is not completely satisfactory yet (duplicate code with
461 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
464 - Simplify transport plugin (internal) interface by not passing any
465 argument to _server and _client, but embedding them in the socket
468 2004-09-28 Martin Quinson
469 - Finish the port to AIX.
470 autoconf was my problem (segfault within the malloc replacement
471 function. No idea why)
473 2004-09-16 Martin Quinson
474 - Fix some size_t madness on 64bit architectures.
476 2004-09-08 Martin Quinson
477 - Reduce the number of system headers loaded, overload some more system
478 calls (such as malloc to cast the result of the system one, and work
480 - Fix and reintroduce the config support
482 2004-09-07 Martin Quinson
483 - Source code reorganization to allow Arnaud to surf all over there.
484 - Allow to document the logging categories.
485 - Remove all uppercase from logging categories and useless cleanup in names.
487 2004-08-18 Martin Quinson
488 Version 0.6.2 (protocol not changed; API changed)
489 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
490 gras_error_t), and NULL when not found. Functions expecting a msgtype
491 as argument (msg_wait; msg_send) deal with NULL argument by providing a
492 hopefully usefull message.
493 - Portability to prehistoric sparcs again
495 2004-08-17 Martin Quinson
496 Version 0.6.1 (protocol not changed; ABI not changed)
497 - prealloc some buffers to speed things up
499 2004-08-11 Martin Quinson
500 Version 0.6 (protocol not changed; ABI expended)
501 - The parsing macro can deal with the references, provided that you add
502 the relevant annotations (using GRAS_ANNOTE(size,field_name))
504 2004-08-09 Martin Quinson
505 Version 0.5 (protocol not changed; ABI changed)
506 - Allow to off turn the cycle detection code in data exchange at
507 compilation time. It should be at run time, but I'm short of time (and
508 the config stuff is still broken). That way, we keep dict out of the
509 critical path, which is good because the performance is poor:
510 - search not dichotomial yet
511 - dynar give no way to access their content and memcpy everytime
512 - In composed data description (struct, ref and so on), stop foolness of
513 keeping the subtype's ID, but store the type itself. This keeps sets out
514 of the critical path, which is good since they rely on dynar and
515 dictionnaries. The only loose of that is that we cannot detect the
516 redeclaration of a structure/union with another content (but I'm not sure
517 the code detected well this error before anyway). We still can detect
518 the redefinition discrepancy for the other types.
519 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
520 it breaks the code because of type-punning used all over the place).
521 This breaks on all non-gcc architectures (for now).
523 All those changes (plus the buffer of last time) allow me to gain 2 order
524 of magnitude on cruel tests consisting of 800000 array of integers on two
525 level of a hierarchical structure (200 secondes -> 4 secondes)
528 - the selector of reference must now return the type it points to, not
531 2004-08-06 Martin Quinson
532 Version 0.4 (protocol changed; ABI not changed)
533 - Allow to pass --gras-log argument to processes in simulation mode. Really.
534 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
535 - Implement a buffer transport, and use it by default (it relies on tcp in
536 real life and on sg in simulation).
537 That's a bit hackish since I had a new field to the structure to store
538 its data without interfering with the subtype ones. Inheritance
539 is tricky in C. And that's a kind of reverse inheritance with one class
540 derivating two classes. Or maybe a game with java interfaces. Anyway,
541 that's damn hard in C (at least).
542 Moreover, I got tired while trying to ensure plugin separation and
543 genericity in SG mode. MSG wants me to do weird things, so let's go for
544 cruel hacks (temporarily of course ;).
545 See comment in transport_private.h:71
546 - do not include all the _interface headers in private but in the files
547 which really need them (to cut the compilation time when they are
550 2004-07-26 Martin Quinson
551 Version 0.3 (protocol not changed; ABI changed)
552 - Major overhault of the datadesc interface to simplify it:
553 - shorted the function names:
554 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
555 - add a trivial way to push/pop integers into the cbps without malloc.
556 This allows to make really generic sub_type description, which simply
557 pop their size of the stack.
558 - add a function gras_datadesc_ref_pop_arr() which does what users want
559 most of the time: Declare a dynamic array (which pops its size of the
560 stack) and declare a reference to it. Poor name, but anyway.
561 - kill the post-send callback, add a post-receive one
563 2004-07-23 Martin Quinson
564 Version 0.2 (protocol changed; ABI changed)
565 - add some testing for cpbs in the test cases, and fix some more bugs.
566 This invalidate again the little64 data file, since I cannot regenerate
568 - remove an awfull optimization in the logging stuff, allowing me to:
569 - understand it again
570 - learn gcc how to check that the argument match the provided format
571 - fix all errors revealed by gcc after that
572 - internal keys of dict are not \0 terminated. Deal with it properly in
573 loggings instead of segfaulting when the user want to see the logs :-/
575 2004-07-22 Martin Quinson
576 - Fix some stupid bug preventing cbps (callback postit) from working
578 2004-07-21 Martin Quinson
579 - Some documentation cleanups
580 - remove the useless last argument of msgtype_declare
581 - rename the Virtu functions to fit into the 'os' namespace
582 - move headers src/include -> src/include/gras/ and stop fooling with
583 gras -> . symbolic link
584 - make distcheck is now successful
586 2004-07-19 Martin Quinson
588 - Build shared library also
589 - Install html doc to the right location
590 - stop removing maintainer files in make clean
591 - build tests only on make check
593 2004-07-13 Martin Quinson
595 - No major issue in previous version => change versionning schema
596 - Re-enable little64 convertion test now that Abdou kindly regenerated the
597 corresponding dataset.
599 2004-07-11 Martin Quinson
601 - Get it working with any kind of structure (we can compute the padding
602 bytes remotely for all the architectures I have access to)
603 - Implement the structure parsing macro (still not quite robust/complete)
604 - Improvement to the remote testing toysuite
606 2004-07-10 Martin Quinson
608 - get ride of a bunch of deprecated macros
609 - actually run the test for two-compliment, not only compile it :-/
610 - test whether the structures get packed (and bail out if yes. Damn.
611 Alignment is a serious matter)
612 - test whether the structures get compacted (but respecting the alignment
613 constraints of each types)
614 - test whether the array fields of structures can straddle alignment boundaries
616 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
617 (took me 2 hours to find that bug, looking at the wrong place)
618 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
619 before sending/receiving and are used to compute the offsets of fields
620 - Given that padding size depend even on compiler options, keep track of
621 alignment and aligned_size only for the current architecture. Not a big
622 deal since we send structure fields one after the other (seems
624 - Add the datastructure used for IEEE paper by the PBIO guys to the test
625 program, let it work on linux/gcc/little32. portability todo.
627 2004-07-08 Martin Quinson
628 - import and improve remote compilation support from FAST
629 - make sure make check works on half a dozen of machines out there
631 2004-07-07 Martin Quinson
632 Let's say it's version 0.0.3 ;)
633 - Implement conversions (yuhu!)
634 - Let it work on solaris (beside conversion, of course)
635 - Stupid me, using rand() to generate the conversion datatests in not wise.
637 2004-07-06 Martin Quinson
638 - Let make dist work, since I'm gonna need it to compile on remote hosts
639 - Let Tests/datadesc_usage write the architecture on which the file was
640 generated as first byte.
641 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
642 architecture descriptions.
643 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
644 on those architectures.
645 - Optimization: send/recv array of scalar in one shoot
647 2004-07-05 Martin Quinson
648 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
650 - Plug a whole bunch of memleaks
651 - each process now have to call gras_{init,exit}. One day, their log
652 settings will be separated
653 - Continue the code factorisation between SG, RL and common in Transport.
655 2004-07-04 Martin Quinson
657 - Redistribution between SG and RL.
658 We wanna have to accept in SG, so move accepted related parts of RL in
659 the common part. (more precisely, the dynar of all known sockets is no
660 more a static in transport.c, but part of the process_data)
662 [gras_stub_generator]
663 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
665 2004-07-03 Martin Quinson
666 - Create a new log channel tbx containing dict, set, log, dynar (to shut
667 them all up in one shot)
669 - Fix the ugly case of reference to dynamic array.
670 - New (semi-public) function gras_datadesc_size to allow the messaging
671 layer to malloc the needed space for the buffer.
673 - gras_socket_close now expect the socket to close (and not its address to
674 put NULL in it after it). This is because the socket passed to handlers
675 is one of their argument (=> not writable).
677 - propagate the interface cleanup from last week in datadesc, ie remove a
678 superfluous level of indirection. User pass adress of variable
679 containing data (both when sending and receiving), and not of a variable
680 being a pointer to the data. Let's say that I like it better ;)
681 The price for that is constructs like "int msg=*(int*)payload" in
682 handlers, but it's a fine price, IMHO.
684 - Let it work in RL (yuhu)
686 2004-06-21 Martin Quinson
688 - porting SG plugin and SG select to new standards (works almost).
689 - plug memleaks and fix bugs around.
692 - cleanup the prototype of data recv and force users to specify when they
693 want to handle references to objects. Test case working even for cycles.
694 - plug memleaks. Valgrind is perfectly ok with this.
696 2004-06-12 Martin Quinson
698 - cleanup the separation between plugin and main code in plugin creation
700 2004-06-11 Martin Quinson
702 - Reput hook for raw sockets, needed for BW experiments
703 - kill a few lines of dead code
704 [Data description] Interface cleanup
705 - gras_datadesc_by_name returns the searched type or NULL.
706 That way, no variable is needed to use a type desc once, which makes
708 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
709 two parameters were strings (field name, type name), leading to
711 [Dicos] Interface cleanup
712 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
713 This is consistant with the dynar API.
715 2004-04-21 Martin Quinson
717 - Porting to new standards.
720 There is no bag anymore, no need to take extra provision to mask the
721 pointers behind "ID".
722 Better splitup of functions between files create/exchange/convert.
723 This is still a bit artificial since convert and receive are so
724 interleaved, but anyway.
726 - add a queued message list to procdata (the ones not matching criteria
728 - factorize some more code between SG and RL wrt procdata
730 - use gras_exit in example to track memleaks
731 - get rid of gs_example now that GS is properly integrated into gras
732 - update run_test to integrate the lastest tests (datadesc)
734 - rename WARNINGn macros to WARNn since it prooved error-prone
736 2004-04-19 Martin Quinson
738 - register init/exit functions within gras module mechanism
739 - send/receive function.
740 Convertion is not implemented, but short-cutted if not needed.
741 struct/array elements are sent one by one (instead of block-wise), but
742 nobody really cares (yet). Get a prototype before optimizing.
743 - tests (using a file socket) for DD send/receive on:
744 - base types: int, float
745 - array: fixed size, string (ie ref to dynamic string)
746 - structure: homogeneous, heterogeneous
747 - chained list, graph with cycle
748 Believe it or not, valgrind is not too unhappy with the results. The
749 cycle happily segfaults, but the others are ok. And I'm sick of pointers
753 - Bugfix when using a filename explicitely (instead of '-')
755 2004-04-09 Martin Quinson
757 - factorize more code between RL and SG in socket creation
758 - Complete the implementation and tests of:
760 o file (only in RL, and mainly for debugging)
762 I lost 3 days to design a portable address resolver, and then decided
763 that the prototype mainly have to run on my box.
764 Addressing portability too early may be like optimizing too early :-/
766 - use gras_init in the Tests instead of the crappy parse_log_opt
767 (the latter function is removed)
768 [Conditional execution]
769 - New functions: gras_if_RL/gras_if_SG (basic support for this)
770 [Code reorganisation]
771 - Get rid of libgrasutils.a since it makes more trouble than it solves.
772 Build examples against the RL library, since there is no way to disable
773 its creation for now.