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]
26 * Use gras_datadesc_copy() to exchange messages on top of SG [MQ]
28 - improve message exchange performance on top of SG
29 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
31 Documentation improvements:
32 * xbt_log module unmercifully reworked [MQ]
36 SimGrid (3.00) stable; urgency=low
39 * New! Give the possibility to hijack the surf parser and thus bypass
40 MSG_create_environment and MSG_launch_application. Have a look at
41 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
43 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
45 SimGrid (2.96) unstable; urgency=low
50 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
51 This deprecates the xbt_error_t mecanisms.
52 It modifies (simplifies) all XBT and GRAS API.
53 MSG API keeps unchanged (exceptions raised by XBT are catched from
54 within MSG and masked with existing error handling facilities)
57 * New! Add a FATPIPE model. [AL]
58 * New! Add a parallel task model. [AL]
59 * New! Add automatically a loopback interface (in the default
60 network model) if none was precised.
63 * Bugfix: MSG_process_resume now works with the current running process.
65 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
66 * Modification of MSG_task_get_compute_duration. Once a task has been
67 processed, the value returned by this function is now equal to 0. [AL]
68 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
69 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
70 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
71 of a m_task. For now, MSG_task_cancel only works with computation
73 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
74 of the processor (in Mflop/s) regardless of the current load on the
76 * API Change: use proper naming convention for MSG_getClock and
77 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
79 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
80 This function changes the priority of a computation task. This priority
81 doesn't affect the transfer rate. A priority of 2 will make a task
82 receive two times more cpu power than the other ones. This function
83 has been added to suit the needs of Nguyen The Loc and hasn't been that
84 much tested yet. So if it fails, please report it and send me your code.
86 * API Change: removed all functions and types that were marked "deprecated"
87 since many months. Renamed MSG_global_init_args to MSG_global_init.
89 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
91 SimGrid (2.95) unstable; urgency=low
94 * Steal some nice code to GNU pth to fix context detection and usage [AL]
95 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
96 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
99 * Bugfix: Allow absolute paths for platform description files [MQ]
100 * Bugfix: do free the variables after use. Leads to drastic performance
102 * Implement max_duration (ie, timeouts) on resources [AL]
105 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
107 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
108 the needed support (timeouts) [AL]
111 * Implement measurement sockets. You can now get the bandwidth between two
112 hosts thanks to AMOK (see below). [MQ]
113 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
114 dynar over the network (yeah) [MQ]
115 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
116 * simple/static token-ring example. [Alexandre Colucci and MQ]
117 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
121 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
122 type parsing. "short" and "long" modifiers were also ignored; other
123 modifier (such as reference level) are still ignored. [MQ]
124 * Embeed the buffer size within the buffer itself on SG. [MQ]
125 That way, send() are atomic and cannot get intermixed anymore (at least
126 the ones which are less than 100k; bigger messages still have the issue)
127 * Array size pushed by the field, not by the field type (or each
128 and every long int will push stuff to the cbps) [MQ]
129 * use select() to sleep since it allows to portably sleep less than one
132 GRAS (minor cleanups)
133 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
136 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
137 * type category 'ignored' killed as it was never used and were difficult
139 * whether a type can cycle or not is now a flag, leaving room for more
140 flags (as "ignored", if we feel the need one day ;)
141 * Rename raw sockets to measurement sockets since "raw" has another
142 meaning in networking community.
145 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
146 and offers features not belonging to GRAS but that most applications
147 need. One day, it may be a set of plugins loadable at runtime.
148 * New module: bandwidth
149 bandwidth measurement between arbitrary nodes running this module. [MQ]
151 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
153 SimGrid (2.94) unstable; urgency=low
155 The first beta release of SimGrid 3 !
159 * Update the main page and the FAQ. Adding references to gforge.
162 * Add a gras_os_getpid function.
165 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
166 * Extend the logs so that they also print PID, hostname, date, ... if
168 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
169 and kill the old version which were in testsuite/
170 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
174 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
175 for users. Instead, each programming environment (GRAS, MSG,...) use it
176 when needed to provide such a feature to users.
177 Don't shortcut the mecanism or you will also shortcut the virtualization
178 you need on the simulator.
182 * Cleanups in configury with regard to compile optimization/warning flags.
183 Also add -fno-loop-optimize to any powerpc since it's the optimization
184 killing gcc (< 3.4.0).
185 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
186 needed by MSG examples complications
187 * Borrow configury beautifications from PHP
190 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
191 warning (thanks loris for stumbling into this one).
192 * Bugfix: stop loading private headers (gras_config.h) from the public
196 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
197 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
198 RL would imply a DNS resolver, which is *hard* to do in a portable way
199 (and therefore delayed).
200 * Implement a real timer mecanism and use it in timing macros. This allows
201 to avoid rounding errors and get a 0.000005 sec precision in timing
202 macros. While I was at it, various cleanups:
203 - allow to declare more than one timed section per file (fix a stupid bug)
204 - move some private declaration to the right place
205 - merge conditional execution and timing macros into emulation module
206 - document the module
207 - make sure the module cleanups its mess on gras_exit
208 * Documentation improvements:
209 - (new) how to compile applications using GRAS
210 - (new) emulation support (timing macros)
212 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
214 SimGrid (2.93) unstable; urgency=low
216 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
219 - Use Paje properly where used. Still to be sanitized properly.
220 - Portability fix: Add an implementation of the contexts using pthread
224 - Add xbt_procname(): returns the name of the current process.
225 Use it to show the current process's name in all logging.
227 - fix detection of older flex version and the reaction, since we do
228 depend on modern ones (we use lex_destroy)
229 - Better separation of SG and RL in the libs: remove all simulation code
230 from libgras. As a result, this lib is now only 200k when stripped.
231 Some of the xbt modules may also be duplicated (two sets and such) and
232 should be cleaned/killed before SG3.
233 - Insist on using xlC on AIX because of weird problems involving gcc there.
234 - Cleanup the make remote stuff. This is now done by scripts
235 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
236 mainly for our private use, but we're working on changing them to user
239 - Bugfix: flush the socket on close only if there is some *output*.
240 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
242 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
245 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
247 SimGrid (2.92) unstable; urgency=low
249 Alpha 3 on the path to SimGrid 3
253 - New! First try of benchmarking macros.
254 - New! First try so that gras_stub_generator generate deployment and
255 remote compilation helpers.
257 - Bugfix: Initialization fix in msg_test.
261 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
263 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
265 - API Change: gras_os_sleep to take the amount of seconds as a double.
266 Accepting an int was error prone since it was the only location where
267 seconds were coded as such. It leaded to damn rounding errors.
268 - Bugfix: Hard to belive that timers ever worked before this.
270 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
272 SimGrid (2.91) unstable; urgency=low
274 Alpha 2 on the path to SimGrid 3
278 - Bug fix in the lmm_solver.
280 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
281 through the function MSG_paje_output.
282 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
283 - It is possible to bound the rate of a communication in MSG with
284 MSG_task_put_bounded() (was already in the previous version but I had forgotten
285 to write it in the changelog).
286 - Bug fix to let GRAS run on top of MSG until we move it directly on top
291 - Various cleanups to the autotools stuff
292 - Begin to move Gras examples to examples/gras/
293 - Let make distcheck work again (yeah!)
295 - documentation overhauled using doxygen.
296 gtk-doc-tools is dead in SimGrid now.
297 - Automatically extract all existing logging categories, and add the list
298 to the documentation (long standing one, to say the less)
300 - Cleanup the known architecture table. Reorder the entries to group what
301 should be, and use a more consistent naming scheme.
302 (some of the test dataset are still to be regenerated)
303 - New! Allow library to register globals on each process just as userdata
305 This is implemented using a xbt_dict and not a xbt_set, so we loose the
306 lookup time (for now).
307 Use it in msg and trp.
308 This cleans a lot the internals and helps enforcing privacy of the
309 headers between the gras components.
310 - New! Add a timer mechanism, not unlike cron(8) and at(1).
311 - Bugfix: gras_os_time was delirious in RL.
312 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
314 - Reenable GRAS now that it works.
316 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
318 SimGrid (2.90) unstable; urgency=low
320 Alpha 1 on the path to SimGrid 3
322 * It is a long time since the last release of SimGrid. I'm sorry about
323 that but as I had told you, I was rewriting a lot of things. I apologize
324 to those who had been reporting bugs to me and that I had not answered.
325 If your bug is still in the new version, please tell me. Here is a
326 summary of the main changes.
328 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
329 lead by Martin Quinson. As a consequence SimGrid gains a lot in
330 portability, speed, and a lot more but you'll figure it out later.
331 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
332 I wanted to release the new MSG as soon as possible and I have
333 broken GRAS, which is the reason why, for now, only MSG is fully
334 functional. A laconic description of these projects is available
335 in the documentation.
337 * REVOLUTION 2: I have removed SG and I am now using a new simulation
338 kernel optimized for our needs (called SURF but only the developers
339 should use it). Hence, MSG is now roughly 30 times faster and I think
340 that by rewriting a little bit MSG, I could event speed it up a little
341 bit more. Beside the gain in speed, it is also much easier to encode a
342 new platform model with SURF than it was with SG. More to come...
344 * REVOLUTION 3: I have tried to change a little as possible the API of
345 MSG but a few things really had to disappear. The main differences
346 with the previous version are :
347 1) no more m_links_t and the corresponding functions. Platforms are
348 directly read from a XML description and cannot be hard-coded
349 anymore. The same format is used for application deployment
350 description. The new format is described in the documentation.
351 Have a look in tools/platform_generation. There is a tiny script
352 that converts from the old platform format to the new one. Concerning
353 the application deployment format, parsing the old one is tricky.
354 I think most of you should however be able to convert your files. If
355 it is really an issue, I can write a C code that does the conversion.
357 2) the toolbox tbx does not exist anymore. We now have a library
358 with much more data-structures but without the hash-tables (we have
359 dictionaries that are much faster).
361 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
363 *****************************************************************************
364 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
365 * don't feel like converting the oldies. (Mt) *
366 *****************************************************************************
369 Version 2.90: "the long awaited one"
370 - Finished rewriting and debugging MSG. Rewrote the documentation.
371 - disable GRAS for now since it needs to be ported to the newest SG
374 - Finish the port to windows (using mingw32 for cross-compile)
377 - Main loop and datastructures of SURF. A cpu resource object is
378 functional. Surf can thus be used to create cpu's with variable
379 performance on which you can execute some actions.
381 2004-11-15 Martin Quinson
382 - Port to ARM. Simply added the alignment and size descriptions. Should
383 work, but the ARM machines are so slow that I didn't had the opportunity
384 to 'make check' over there yet.
386 2004-11-15 Arnaud Legrand
387 - Trace manager now written. It uses a heap structure and is therefore
388 expected to be efficient. It may however be speeded up (particularly
389 when many events occur at the same date) by using red and black
390 trees. One day maybe...
391 - Max-min linear system solver written. It uses a sparse matrix
392 structure taking advantage of its expected use. Most operations are
393 O(1) and free/calloc are called as few as possible. The computation of
394 the minimum could however be improved by using a red and black tree
397 2004-11-03 Arnaud Legrand
398 - Rename every gras_* function that was in xbt/ to its xbt_
400 - Add a heap and a doubly-linked list to xbt
401 - Added a dichotomy to the dictionaries. make check works as well before
402 so I assume that the patch is correct. I do not know however if things
403 run effectively faster than before now. :)
405 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
406 SimGrid, and the version number is bumped to 2.x
408 2004-10-29 Martin Quinson
409 - Introduction of the remote errors.
410 They are the result of a RMI/RPC on the remote machine.
411 ErrCodes being scalar values, you can't get the host on which those
412 errors did happen. Extending the error mechanism as in Gnome is possible.
413 No idea yet whether it is a good idea.
415 2004-10-28 Martin Quinson
416 - Interface revolution: the Starred Structure Eradication.
417 I used to do typedef struct {} toto_t; and then handle *toto_t.
418 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
419 - ???_t is a valid type (builded with typedef)
420 - s_toto_t is a structure (access to fields with .)
421 - s_toto is a structure needing 'struct' keyword to be used
422 - e_toto_t is an enum
423 - toto_t is an 'object' (struct*)
425 typedef struct s_toto {} s_toto_t, *toto_t;
426 typedef enum {} e_toto_t;
427 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
428 s_toto_t) is private.
430 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
431 so that it can set the variable to NULL. It was so for dicts and sets,
432 it changed for dynars.
434 - Fix a bunch of memleaks in dict_remove
435 - Fix a bug in sg/server_socket opening: it failed all the time.
437 2004-10-07 Martin Quinson
438 - Speed up dynar lookup operation a bit.
440 gras_dynar_get is dead.
442 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
443 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
444 (which gives you the address of the stored data).
446 gras_dynar_get_as is an helpful macro which allows you to retrieve a
447 copy of the data using an affectation to do the job and not a memcpy.
449 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
450 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
452 It does not really speedup the dynar test because they are
453 setting elements all the time (and look them seldom). But the dict does
454 far more lookup than setting.
456 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
458 2004-10-05 Martin Quinson
459 - Allow to (en/dis)able the cycle detection at run time.
461 Whether we should check for cycle or not is now a property of each
462 datatype. When you think there may be some cycle, use datadesc_cycle_set.
463 datadesc_cycle_unset allow to remove this property when previously set.
465 Note that the cycle detection is off by default since it impacts the
466 performance. Watch the data you feed GRAS with ;)
468 This property is hereditary. Any element embedded in a structure having it
469 set have it set for the time of this data exchange.
471 You should set it both on sender and receiver side. If you don't set it on
472 sender side, it will enter an endless loop. If you forget on receiver
473 side, the cycles won't be recreated after communication.
475 - Header reorganization.
476 Kill gras_private.h, each submodule must load the headers it needs.
478 2004-10-04 Martin Quinson
479 - Interface revolution: do not try to survive to malloc failure.
481 Now, gras_malloc and friends call gras_abort() on failure.
482 As a conclusion, malloc_error is not a valid error anymore, and all
483 functions for which it was the only gras_error_t return value are
484 changed. They now return void, or there result directly.
485 This simplify the API a lot.
487 2004-09-29 Martin Quinson
488 - Re-enable raw sockets.
489 Created by gras_socket_{client,server}_ext;
490 Used with gras_raw_{send,recv}
493 It should allow to kill the last bits of gras first version soon.
495 This is not completely satisfactory yet (duplicate code with
496 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
499 - Simplify transport plugin (internal) interface by not passing any
500 argument to _server and _client, but embedding them in the socket
503 2004-09-28 Martin Quinson
504 - Finish the port to AIX.
505 autoconf was my problem (segfault within the malloc replacement
506 function. No idea why)
508 2004-09-16 Martin Quinson
509 - Fix some size_t madness on 64bit architectures.
511 2004-09-08 Martin Quinson
512 - Reduce the number of system headers loaded, overload some more system
513 calls (such as malloc to cast the result of the system one, and work
515 - Fix and reintroduce the config support
517 2004-09-07 Martin Quinson
518 - Source code reorganization to allow Arnaud to surf all over there.
519 - Allow to document the logging categories.
520 - Remove all uppercase from logging categories and useless cleanup in names.
522 2004-08-18 Martin Quinson
523 Version 0.6.2 (protocol not changed; API changed)
524 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
525 gras_error_t), and NULL when not found. Functions expecting a msgtype
526 as argument (msg_wait; msg_send) deal with NULL argument by providing a
527 hopefully usefull message.
528 - Portability to prehistoric sparcs again
530 2004-08-17 Martin Quinson
531 Version 0.6.1 (protocol not changed; ABI not changed)
532 - prealloc some buffers to speed things up
534 2004-08-11 Martin Quinson
535 Version 0.6 (protocol not changed; ABI expended)
536 - The parsing macro can deal with the references, provided that you add
537 the relevant annotations (using GRAS_ANNOTE(size,field_name))
539 2004-08-09 Martin Quinson
540 Version 0.5 (protocol not changed; ABI changed)
541 - Allow to off turn the cycle detection code in data exchange at
542 compilation time. It should be at run time, but I'm short of time (and
543 the config stuff is still broken). That way, we keep dict out of the
544 critical path, which is good because the performance is poor:
545 - search not dichotomial yet
546 - dynar give no way to access their content and memcpy everytime
547 - In composed data description (struct, ref and so on), stop foolness of
548 keeping the subtype's ID, but store the type itself. This keeps sets out
549 of the critical path, which is good since they rely on dynar and
550 dictionnaries. The only loose of that is that we cannot detect the
551 redeclaration of a structure/union with another content (but I'm not sure
552 the code detected well this error before anyway). We still can detect
553 the redefinition discrepancy for the other types.
554 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
555 it breaks the code because of type-punning used all over the place).
556 This breaks on all non-gcc architectures (for now).
558 All those changes (plus the buffer of last time) allow me to gain 2 order
559 of magnitude on cruel tests consisting of 800000 array of integers on two
560 level of a hierarchical structure (200 secondes -> 4 secondes)
563 - the selector of reference must now return the type it points to, not
566 2004-08-06 Martin Quinson
567 Version 0.4 (protocol changed; ABI not changed)
568 - Allow to pass --gras-log argument to processes in simulation mode. Really.
569 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
570 - Implement a buffer transport, and use it by default (it relies on tcp in
571 real life and on sg in simulation).
572 That's a bit hackish since I had a new field to the structure to store
573 its data without interfering with the subtype ones. Inheritance
574 is tricky in C. And that's a kind of reverse inheritance with one class
575 derivating two classes. Or maybe a game with java interfaces. Anyway,
576 that's damn hard in C (at least).
577 Moreover, I got tired while trying to ensure plugin separation and
578 genericity in SG mode. MSG wants me to do weird things, so let's go for
579 cruel hacks (temporarily of course ;).
580 See comment in transport_private.h:71
581 - do not include all the _interface headers in private but in the files
582 which really need them (to cut the compilation time when they are
585 2004-07-26 Martin Quinson
586 Version 0.3 (protocol not changed; ABI changed)
587 - Major overhault of the datadesc interface to simplify it:
588 - shorted the function names:
589 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
590 - add a trivial way to push/pop integers into the cbps without malloc.
591 This allows to make really generic sub_type description, which simply
592 pop their size of the stack.
593 - add a function gras_datadesc_ref_pop_arr() which does what users want
594 most of the time: Declare a dynamic array (which pops its size of the
595 stack) and declare a reference to it. Poor name, but anyway.
596 - kill the post-send callback, add a post-receive one
598 2004-07-23 Martin Quinson
599 Version 0.2 (protocol changed; ABI changed)
600 - add some testing for cpbs in the test cases, and fix some more bugs.
601 This invalidate again the little64 data file, since I cannot regenerate
603 - remove an awfull optimization in the logging stuff, allowing me to:
604 - understand it again
605 - learn gcc how to check that the argument match the provided format
606 - fix all errors revealed by gcc after that
607 - internal keys of dict are not \0 terminated. Deal with it properly in
608 loggings instead of segfaulting when the user want to see the logs :-/
610 2004-07-22 Martin Quinson
611 - Fix some stupid bug preventing cbps (callback postit) from working
613 2004-07-21 Martin Quinson
614 - Some documentation cleanups
615 - remove the useless last argument of msgtype_declare
616 - rename the Virtu functions to fit into the 'os' namespace
617 - move headers src/include -> src/include/gras/ and stop fooling with
618 gras -> . symbolic link
619 - make distcheck is now successful
621 2004-07-19 Martin Quinson
623 - Build shared library also
624 - Install html doc to the right location
625 - stop removing maintainer files in make clean
626 - build tests only on make check
628 2004-07-13 Martin Quinson
630 - No major issue in previous version => change versionning schema
631 - Re-enable little64 convertion test now that Abdou kindly regenerated the
632 corresponding dataset.
634 2004-07-11 Martin Quinson
636 - Get it working with any kind of structure (we can compute the padding
637 bytes remotely for all the architectures I have access to)
638 - Implement the structure parsing macro (still not quite robust/complete)
639 - Improvement to the remote testing toysuite
641 2004-07-10 Martin Quinson
643 - get ride of a bunch of deprecated macros
644 - actually run the test for two-compliment, not only compile it :-/
645 - test whether the structures get packed (and bail out if yes. Damn.
646 Alignment is a serious matter)
647 - test whether the structures get compacted (but respecting the alignment
648 constraints of each types)
649 - test whether the array fields of structures can straddle alignment boundaries
651 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
652 (took me 2 hours to find that bug, looking at the wrong place)
653 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
654 before sending/receiving and are used to compute the offsets of fields
655 - Given that padding size depend even on compiler options, keep track of
656 alignment and aligned_size only for the current architecture. Not a big
657 deal since we send structure fields one after the other (seems
659 - Add the datastructure used for IEEE paper by the PBIO guys to the test
660 program, let it work on linux/gcc/little32. portability todo.
662 2004-07-08 Martin Quinson
663 - import and improve remote compilation support from FAST
664 - make sure make check works on half a dozen of machines out there
666 2004-07-07 Martin Quinson
667 Let's say it's version 0.0.3 ;)
668 - Implement conversions (yuhu!)
669 - Let it work on solaris (beside conversion, of course)
670 - Stupid me, using rand() to generate the conversion datatests in not wise.
672 2004-07-06 Martin Quinson
673 - Let make dist work, since I'm gonna need it to compile on remote hosts
674 - Let Tests/datadesc_usage write the architecture on which the file was
675 generated as first byte.
676 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
677 architecture descriptions.
678 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
679 on those architectures.
680 - Optimization: send/recv array of scalar in one shoot
682 2004-07-05 Martin Quinson
683 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
685 - Plug a whole bunch of memleaks
686 - each process now have to call gras_{init,exit}. One day, their log
687 settings will be separated
688 - Continue the code factorisation between SG, RL and common in Transport.
690 2004-07-04 Martin Quinson
692 - Redistribution between SG and RL.
693 We wanna have to accept in SG, so move accepted related parts of RL in
694 the common part. (more precisely, the dynar of all known sockets is no
695 more a static in transport.c, but part of the process_data)
697 [gras_stub_generator]
698 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
700 2004-07-03 Martin Quinson
701 - Create a new log channel tbx containing dict, set, log, dynar (to shut
702 them all up in one shot)
704 - Fix the ugly case of reference to dynamic array.
705 - New (semi-public) function gras_datadesc_size to allow the messaging
706 layer to malloc the needed space for the buffer.
708 - gras_socket_close now expect the socket to close (and not its address to
709 put NULL in it after it). This is because the socket passed to handlers
710 is one of their argument (=> not writable).
712 - propagate the interface cleanup from last week in datadesc, ie remove a
713 superfluous level of indirection. User pass adress of variable
714 containing data (both when sending and receiving), and not of a variable
715 being a pointer to the data. Let's say that I like it better ;)
716 The price for that is constructs like "int msg=*(int*)payload" in
717 handlers, but it's a fine price, IMHO.
719 - Let it work in RL (yuhu)
721 2004-06-21 Martin Quinson
723 - porting SG plugin and SG select to new standards (works almost).
724 - plug memleaks and fix bugs around.
727 - cleanup the prototype of data recv and force users to specify when they
728 want to handle references to objects. Test case working even for cycles.
729 - plug memleaks. Valgrind is perfectly ok with this.
731 2004-06-12 Martin Quinson
733 - cleanup the separation between plugin and main code in plugin creation
735 2004-06-11 Martin Quinson
737 - Reput hook for raw sockets, needed for BW experiments
738 - kill a few lines of dead code
739 [Data description] Interface cleanup
740 - gras_datadesc_by_name returns the searched type or NULL.
741 That way, no variable is needed to use a type desc once, which makes
743 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
744 two parameters were strings (field name, type name), leading to
746 [Dicos] Interface cleanup
747 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
748 This is consistant with the dynar API.
750 2004-04-21 Martin Quinson
752 - Porting to new standards.
755 There is no bag anymore, no need to take extra provision to mask the
756 pointers behind "ID".
757 Better splitup of functions between files create/exchange/convert.
758 This is still a bit artificial since convert and receive are so
759 interleaved, but anyway.
761 - add a queued message list to procdata (the ones not matching criteria
763 - factorize some more code between SG and RL wrt procdata
765 - use gras_exit in example to track memleaks
766 - get rid of gs_example now that GS is properly integrated into gras
767 - update run_test to integrate the lastest tests (datadesc)
769 - rename WARNINGn macros to WARNn since it prooved error-prone
771 2004-04-19 Martin Quinson
773 - register init/exit functions within gras module mechanism
774 - send/receive function.
775 Convertion is not implemented, but short-cutted if not needed.
776 struct/array elements are sent one by one (instead of block-wise), but
777 nobody really cares (yet). Get a prototype before optimizing.
778 - tests (using a file socket) for DD send/receive on:
779 - base types: int, float
780 - array: fixed size, string (ie ref to dynamic string)
781 - structure: homogeneous, heterogeneous
782 - chained list, graph with cycle
783 Believe it or not, valgrind is not too unhappy with the results. The
784 cycle happily segfaults, but the others are ok. And I'm sick of pointers
788 - Bugfix when using a filename explicitely (instead of '-')
790 2004-04-09 Martin Quinson
792 - factorize more code between RL and SG in socket creation
793 - Complete the implementation and tests of:
795 o file (only in RL, and mainly for debugging)
797 I lost 3 days to design a portable address resolver, and then decided
798 that the prototype mainly have to run on my box.
799 Addressing portability too early may be like optimizing too early :-/
801 - use gras_init in the Tests instead of the crappy parse_log_opt
802 (the latter function is removed)
803 [Conditional execution]
804 - New functions: gras_if_RL/gras_if_SG (basic support for this)
805 [Code reorganisation]
806 - Get rid of libgrasutils.a since it makes more trouble than it solves.
807 Build examples against the RL library, since there is no way to disable
808 its creation for now.