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]
18 * Documentation: xbt_log module unmercifully reworked [MQ]
19 * Bugfix: there was a problem with the ending of contexts with
20 the pthread backend. It caused some weird deadlock or behavior
21 depending on the pthread implementation. [AL]
24 * Add additionnal checkings on channel values in communication [AL]
25 * New: MSG_task_get_source to see on which host a task was generated [HC]
27 GRAS performance improvements: [MQ]
29 * Reduce the amount of cbps creation/destruction by making it static to
30 datadesc_send/recv() and using a (newly created) cbps_reset (based on
33 * Change libdata to a set so that we can search for stuff by ID (and thus
34 reduce the insane amount of dict lookups)
37 * Actually implement gras_datadesc_copy() so that we don't have to mimick
38 RL communication on top of SG since it's so uneffective.
39 It may also be used for inter-thread communication in RL, one day.
40 * Use gras_datadesc_copy() to exchange messages on top of SG
42 - improve message exchange performance on top of SG
43 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
44 * Don't exchange on the network the size of the used part of buffer,
45 instead, specify the possible buffer size to read().
47 - reduces the amount of read/write calls (one pair per exchange)
48 - reduces the amount of exchanged data (the size)
49 - allows to retrieve all arrived data on receiver side, if we don't need
50 it right now (subsequent read will peek the buffer)
51 - allows the receiver to proceed with the begining of the stream before
53 - make it possible to build an iov transport (using readv/writev)
55 - take care of the data with non-stable storage (like stacked data),
57 * If possible, TCP send uses vector I/O (when writev() is here)
58 - Don't use it for receive since we send data sizes and data on the
59 same stream, so we wouldn't be able to chain large amount of chunks
60 before having to flush the stuff to read the size.
61 * Rework the transport plugin mecanism to simplify it and reduce the
62 amount of pointer dereferencement when searching for the right function
65 * I guess that now, we do almost as few system calls as possible while
66 doing as few data copy as possible.
68 To improve it further, we could try to send all the sizes first and then
69 all the data (to use iov on receiving size), but it's only a partial
70 solution: when you have 2 dimensional data, the sizes of the second
71 dimension is data of the first dimension, so you need 3 streams.
73 I'm not sure the potential performance gains justify the coding burden.
78 SimGrid (3.00) stable; urgency=low
81 * New! Give the possibility to hijack the surf parser and thus bypass
82 MSG_create_environment and MSG_launch_application. Have a look at
83 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
85 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
87 SimGrid (2.96) unstable; urgency=low
92 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
93 This deprecates the xbt_error_t mecanisms.
94 It modifies (simplifies) all XBT and GRAS API.
95 MSG API keeps unchanged (exceptions raised by XBT are catched from
96 within MSG and masked with existing error handling facilities)
99 * New! Add a FATPIPE model. [AL]
100 * New! Add a parallel task model. [AL]
101 * New! Add automatically a loopback interface (in the default
102 network model) if none was precised.
105 * Bugfix: MSG_process_resume now works with the current running process.
107 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
108 * Modification of MSG_task_get_compute_duration. Once a task has been
109 processed, the value returned by this function is now equal to 0. [AL]
110 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
111 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
112 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
113 of a m_task. For now, MSG_task_cancel only works with computation
115 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
116 of the processor (in Mflop/s) regardless of the current load on the
118 * API Change: use proper naming convention for MSG_getClock and
119 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
121 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
122 This function changes the priority of a computation task. This priority
123 doesn't affect the transfer rate. A priority of 2 will make a task
124 receive two times more cpu power than the other ones. This function
125 has been added to suit the needs of Nguyen The Loc and hasn't been that
126 much tested yet. So if it fails, please report it and send me your code.
128 * API Change: removed all functions and types that were marked "deprecated"
129 since many months. Renamed MSG_global_init_args to MSG_global_init.
131 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
133 SimGrid (2.95) unstable; urgency=low
136 * Steal some nice code to GNU pth to fix context detection and usage [AL]
137 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
138 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
141 * Bugfix: Allow absolute paths for platform description files [MQ]
142 * Bugfix: do free the variables after use. Leads to drastic performance
144 * Implement max_duration (ie, timeouts) on resources [AL]
147 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
149 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
150 the needed support (timeouts) [AL]
153 * Implement measurement sockets. You can now get the bandwidth between two
154 hosts thanks to AMOK (see below). [MQ]
155 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
156 dynar over the network (yeah) [MQ]
157 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
158 * simple/static token-ring example. [Alexandre Colucci and MQ]
159 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
163 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
164 type parsing. "short" and "long" modifiers were also ignored; other
165 modifier (such as reference level) are still ignored. [MQ]
166 * Embeed the buffer size within the buffer itself on SG. [MQ]
167 That way, send() are atomic and cannot get intermixed anymore (at least
168 the ones which are less than 100k; bigger messages still have the issue)
169 * Array size pushed by the field, not by the field type (or each
170 and every long int will push stuff to the cbps) [MQ]
171 * use select() to sleep since it allows to portably sleep less than one
174 GRAS (minor cleanups)
175 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
178 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
179 * type category 'ignored' killed as it was never used and were difficult
181 * whether a type can cycle or not is now a flag, leaving room for more
182 flags (as "ignored", if we feel the need one day ;)
183 * Rename raw sockets to measurement sockets since "raw" has another
184 meaning in networking community.
187 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
188 and offers features not belonging to GRAS but that most applications
189 need. One day, it may be a set of plugins loadable at runtime.
190 * New module: bandwidth
191 bandwidth measurement between arbitrary nodes running this module. [MQ]
193 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
195 SimGrid (2.94) unstable; urgency=low
197 The first beta release of SimGrid 3 !
201 * Update the main page and the FAQ. Adding references to gforge.
204 * Add a gras_os_getpid function.
207 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
208 * Extend the logs so that they also print PID, hostname, date, ... if
210 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
211 and kill the old version which were in testsuite/
212 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
216 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
217 for users. Instead, each programming environment (GRAS, MSG,...) use it
218 when needed to provide such a feature to users.
219 Don't shortcut the mecanism or you will also shortcut the virtualization
220 you need on the simulator.
224 * Cleanups in configury with regard to compile optimization/warning flags.
225 Also add -fno-loop-optimize to any powerpc since it's the optimization
226 killing gcc (< 3.4.0).
227 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
228 needed by MSG examples complications
229 * Borrow configury beautifications from PHP
232 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
233 warning (thanks loris for stumbling into this one).
234 * Bugfix: stop loading private headers (gras_config.h) from the public
238 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
239 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
240 RL would imply a DNS resolver, which is *hard* to do in a portable way
241 (and therefore delayed).
242 * Implement a real timer mecanism and use it in timing macros. This allows
243 to avoid rounding errors and get a 0.000005 sec precision in timing
244 macros. While I was at it, various cleanups:
245 - allow to declare more than one timed section per file (fix a stupid bug)
246 - move some private declaration to the right place
247 - merge conditional execution and timing macros into emulation module
248 - document the module
249 - make sure the module cleanups its mess on gras_exit
250 * Documentation improvements:
251 - (new) how to compile applications using GRAS
252 - (new) emulation support (timing macros)
254 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
256 SimGrid (2.93) unstable; urgency=low
258 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
261 - Use Paje properly where used. Still to be sanitized properly.
262 - Portability fix: Add an implementation of the contexts using pthread
266 - Add xbt_procname(): returns the name of the current process.
267 Use it to show the current process's name in all logging.
269 - fix detection of older flex version and the reaction, since we do
270 depend on modern ones (we use lex_destroy)
271 - Better separation of SG and RL in the libs: remove all simulation code
272 from libgras. As a result, this lib is now only 200k when stripped.
273 Some of the xbt modules may also be duplicated (two sets and such) and
274 should be cleaned/killed before SG3.
275 - Insist on using xlC on AIX because of weird problems involving gcc there.
276 - Cleanup the make remote stuff. This is now done by scripts
277 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
278 mainly for our private use, but we're working on changing them to user
281 - Bugfix: flush the socket on close only if there is some *output*.
282 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
284 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
287 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
289 SimGrid (2.92) unstable; urgency=low
291 Alpha 3 on the path to SimGrid 3
295 - New! First try of benchmarking macros.
296 - New! First try so that gras_stub_generator generate deployment and
297 remote compilation helpers.
299 - Bugfix: Initialization fix in msg_test.
303 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
305 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
307 - API Change: gras_os_sleep to take the amount of seconds as a double.
308 Accepting an int was error prone since it was the only location where
309 seconds were coded as such. It leaded to damn rounding errors.
310 - Bugfix: Hard to belive that timers ever worked before this.
312 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
314 SimGrid (2.91) unstable; urgency=low
316 Alpha 2 on the path to SimGrid 3
320 - Bug fix in the lmm_solver.
322 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
323 through the function MSG_paje_output.
324 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
325 - It is possible to bound the rate of a communication in MSG with
326 MSG_task_put_bounded() (was already in the previous version but I had forgotten
327 to write it in the changelog).
328 - Bug fix to let GRAS run on top of MSG until we move it directly on top
333 - Various cleanups to the autotools stuff
334 - Begin to move Gras examples to examples/gras/
335 - Let make distcheck work again (yeah!)
337 - documentation overhauled using doxygen.
338 gtk-doc-tools is dead in SimGrid now.
339 - Automatically extract all existing logging categories, and add the list
340 to the documentation (long standing one, to say the less)
342 - Cleanup the known architecture table. Reorder the entries to group what
343 should be, and use a more consistent naming scheme.
344 (some of the test dataset are still to be regenerated)
345 - New! Allow library to register globals on each process just as userdata
347 This is implemented using a xbt_dict and not a xbt_set, so we loose the
348 lookup time (for now).
349 Use it in msg and trp.
350 This cleans a lot the internals and helps enforcing privacy of the
351 headers between the gras components.
352 - New! Add a timer mechanism, not unlike cron(8) and at(1).
353 - Bugfix: gras_os_time was delirious in RL.
354 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
356 - Reenable GRAS now that it works.
358 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
360 SimGrid (2.90) unstable; urgency=low
362 Alpha 1 on the path to SimGrid 3
364 * It is a long time since the last release of SimGrid. I'm sorry about
365 that but as I had told you, I was rewriting a lot of things. I apologize
366 to those who had been reporting bugs to me and that I had not answered.
367 If your bug is still in the new version, please tell me. Here is a
368 summary of the main changes.
370 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
371 lead by Martin Quinson. As a consequence SimGrid gains a lot in
372 portability, speed, and a lot more but you'll figure it out later.
373 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
374 I wanted to release the new MSG as soon as possible and I have
375 broken GRAS, which is the reason why, for now, only MSG is fully
376 functional. A laconic description of these projects is available
377 in the documentation.
379 * REVOLUTION 2: I have removed SG and I am now using a new simulation
380 kernel optimized for our needs (called SURF but only the developers
381 should use it). Hence, MSG is now roughly 30 times faster and I think
382 that by rewriting a little bit MSG, I could event speed it up a little
383 bit more. Beside the gain in speed, it is also much easier to encode a
384 new platform model with SURF than it was with SG. More to come...
386 * REVOLUTION 3: I have tried to change a little as possible the API of
387 MSG but a few things really had to disappear. The main differences
388 with the previous version are :
389 1) no more m_links_t and the corresponding functions. Platforms are
390 directly read from a XML description and cannot be hard-coded
391 anymore. The same format is used for application deployment
392 description. The new format is described in the documentation.
393 Have a look in tools/platform_generation. There is a tiny script
394 that converts from the old platform format to the new one. Concerning
395 the application deployment format, parsing the old one is tricky.
396 I think most of you should however be able to convert your files. If
397 it is really an issue, I can write a C code that does the conversion.
399 2) the toolbox tbx does not exist anymore. We now have a library
400 with much more data-structures but without the hash-tables (we have
401 dictionaries that are much faster).
403 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
405 *****************************************************************************
406 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
407 * don't feel like converting the oldies. (Mt) *
408 *****************************************************************************
411 Version 2.90: "the long awaited one"
412 - Finished rewriting and debugging MSG. Rewrote the documentation.
413 - disable GRAS for now since it needs to be ported to the newest SG
416 - Finish the port to windows (using mingw32 for cross-compile)
419 - Main loop and datastructures of SURF. A cpu resource object is
420 functional. Surf can thus be used to create cpu's with variable
421 performance on which you can execute some actions.
423 2004-11-15 Martin Quinson
424 - Port to ARM. Simply added the alignment and size descriptions. Should
425 work, but the ARM machines are so slow that I didn't had the opportunity
426 to 'make check' over there yet.
428 2004-11-15 Arnaud Legrand
429 - Trace manager now written. It uses a heap structure and is therefore
430 expected to be efficient. It may however be speeded up (particularly
431 when many events occur at the same date) by using red and black
432 trees. One day maybe...
433 - Max-min linear system solver written. It uses a sparse matrix
434 structure taking advantage of its expected use. Most operations are
435 O(1) and free/calloc are called as few as possible. The computation of
436 the minimum could however be improved by using a red and black tree
439 2004-11-03 Arnaud Legrand
440 - Rename every gras_* function that was in xbt/ to its xbt_
442 - Add a heap and a doubly-linked list to xbt
443 - Added a dichotomy to the dictionaries. make check works as well before
444 so I assume that the patch is correct. I do not know however if things
445 run effectively faster than before now. :)
447 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
448 SimGrid, and the version number is bumped to 2.x
450 2004-10-29 Martin Quinson
451 - Introduction of the remote errors.
452 They are the result of a RMI/RPC on the remote machine.
453 ErrCodes being scalar values, you can't get the host on which those
454 errors did happen. Extending the error mechanism as in Gnome is possible.
455 No idea yet whether it is a good idea.
457 2004-10-28 Martin Quinson
458 - Interface revolution: the Starred Structure Eradication.
459 I used to do typedef struct {} toto_t; and then handle *toto_t.
460 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
461 - ???_t is a valid type (builded with typedef)
462 - s_toto_t is a structure (access to fields with .)
463 - s_toto is a structure needing 'struct' keyword to be used
464 - e_toto_t is an enum
465 - toto_t is an 'object' (struct*)
467 typedef struct s_toto {} s_toto_t, *toto_t;
468 typedef enum {} e_toto_t;
469 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
470 s_toto_t) is private.
472 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
473 so that it can set the variable to NULL. It was so for dicts and sets,
474 it changed for dynars.
476 - Fix a bunch of memleaks in dict_remove
477 - Fix a bug in sg/server_socket opening: it failed all the time.
479 2004-10-07 Martin Quinson
480 - Speed up dynar lookup operation a bit.
482 gras_dynar_get is dead.
484 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
485 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
486 (which gives you the address of the stored data).
488 gras_dynar_get_as is an helpful macro which allows you to retrieve a
489 copy of the data using an affectation to do the job and not a memcpy.
491 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
492 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
494 It does not really speedup the dynar test because they are
495 setting elements all the time (and look them seldom). But the dict does
496 far more lookup than setting.
498 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
500 2004-10-05 Martin Quinson
501 - Allow to (en/dis)able the cycle detection at run time.
503 Whether we should check for cycle or not is now a property of each
504 datatype. When you think there may be some cycle, use datadesc_cycle_set.
505 datadesc_cycle_unset allow to remove this property when previously set.
507 Note that the cycle detection is off by default since it impacts the
508 performance. Watch the data you feed GRAS with ;)
510 This property is hereditary. Any element embedded in a structure having it
511 set have it set for the time of this data exchange.
513 You should set it both on sender and receiver side. If you don't set it on
514 sender side, it will enter an endless loop. If you forget on receiver
515 side, the cycles won't be recreated after communication.
517 - Header reorganization.
518 Kill gras_private.h, each submodule must load the headers it needs.
520 2004-10-04 Martin Quinson
521 - Interface revolution: do not try to survive to malloc failure.
523 Now, gras_malloc and friends call gras_abort() on failure.
524 As a conclusion, malloc_error is not a valid error anymore, and all
525 functions for which it was the only gras_error_t return value are
526 changed. They now return void, or there result directly.
527 This simplify the API a lot.
529 2004-09-29 Martin Quinson
530 - Re-enable raw sockets.
531 Created by gras_socket_{client,server}_ext;
532 Used with gras_raw_{send,recv}
535 It should allow to kill the last bits of gras first version soon.
537 This is not completely satisfactory yet (duplicate code with
538 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
541 - Simplify transport plugin (internal) interface by not passing any
542 argument to _server and _client, but embedding them in the socket
545 2004-09-28 Martin Quinson
546 - Finish the port to AIX.
547 autoconf was my problem (segfault within the malloc replacement
548 function. No idea why)
550 2004-09-16 Martin Quinson
551 - Fix some size_t madness on 64bit architectures.
553 2004-09-08 Martin Quinson
554 - Reduce the number of system headers loaded, overload some more system
555 calls (such as malloc to cast the result of the system one, and work
557 - Fix and reintroduce the config support
559 2004-09-07 Martin Quinson
560 - Source code reorganization to allow Arnaud to surf all over there.
561 - Allow to document the logging categories.
562 - Remove all uppercase from logging categories and useless cleanup in names.
564 2004-08-18 Martin Quinson
565 Version 0.6.2 (protocol not changed; API changed)
566 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
567 gras_error_t), and NULL when not found. Functions expecting a msgtype
568 as argument (msg_wait; msg_send) deal with NULL argument by providing a
569 hopefully usefull message.
570 - Portability to prehistoric sparcs again
572 2004-08-17 Martin Quinson
573 Version 0.6.1 (protocol not changed; ABI not changed)
574 - prealloc some buffers to speed things up
576 2004-08-11 Martin Quinson
577 Version 0.6 (protocol not changed; ABI expended)
578 - The parsing macro can deal with the references, provided that you add
579 the relevant annotations (using GRAS_ANNOTE(size,field_name))
581 2004-08-09 Martin Quinson
582 Version 0.5 (protocol not changed; ABI changed)
583 - Allow to off turn the cycle detection code in data exchange at
584 compilation time. It should be at run time, but I'm short of time (and
585 the config stuff is still broken). That way, we keep dict out of the
586 critical path, which is good because the performance is poor:
587 - search not dichotomial yet
588 - dynar give no way to access their content and memcpy everytime
589 - In composed data description (struct, ref and so on), stop foolness of
590 keeping the subtype's ID, but store the type itself. This keeps sets out
591 of the critical path, which is good since they rely on dynar and
592 dictionnaries. The only loose of that is that we cannot detect the
593 redeclaration of a structure/union with another content (but I'm not sure
594 the code detected well this error before anyway). We still can detect
595 the redefinition discrepancy for the other types.
596 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
597 it breaks the code because of type-punning used all over the place).
598 This breaks on all non-gcc architectures (for now).
600 All those changes (plus the buffer of last time) allow me to gain 2 order
601 of magnitude on cruel tests consisting of 800000 array of integers on two
602 level of a hierarchical structure (200 secondes -> 4 secondes)
605 - the selector of reference must now return the type it points to, not
608 2004-08-06 Martin Quinson
609 Version 0.4 (protocol changed; ABI not changed)
610 - Allow to pass --gras-log argument to processes in simulation mode. Really.
611 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
612 - Implement a buffer transport, and use it by default (it relies on tcp in
613 real life and on sg in simulation).
614 That's a bit hackish since I had a new field to the structure to store
615 its data without interfering with the subtype ones. Inheritance
616 is tricky in C. And that's a kind of reverse inheritance with one class
617 derivating two classes. Or maybe a game with java interfaces. Anyway,
618 that's damn hard in C (at least).
619 Moreover, I got tired while trying to ensure plugin separation and
620 genericity in SG mode. MSG wants me to do weird things, so let's go for
621 cruel hacks (temporarily of course ;).
622 See comment in transport_private.h:71
623 - do not include all the _interface headers in private but in the files
624 which really need them (to cut the compilation time when they are
627 2004-07-26 Martin Quinson
628 Version 0.3 (protocol not changed; ABI changed)
629 - Major overhault of the datadesc interface to simplify it:
630 - shorted the function names:
631 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
632 - add a trivial way to push/pop integers into the cbps without malloc.
633 This allows to make really generic sub_type description, which simply
634 pop their size of the stack.
635 - add a function gras_datadesc_ref_pop_arr() which does what users want
636 most of the time: Declare a dynamic array (which pops its size of the
637 stack) and declare a reference to it. Poor name, but anyway.
638 - kill the post-send callback, add a post-receive one
640 2004-07-23 Martin Quinson
641 Version 0.2 (protocol changed; ABI changed)
642 - add some testing for cpbs in the test cases, and fix some more bugs.
643 This invalidate again the little64 data file, since I cannot regenerate
645 - remove an awfull optimization in the logging stuff, allowing me to:
646 - understand it again
647 - learn gcc how to check that the argument match the provided format
648 - fix all errors revealed by gcc after that
649 - internal keys of dict are not \0 terminated. Deal with it properly in
650 loggings instead of segfaulting when the user want to see the logs :-/
652 2004-07-22 Martin Quinson
653 - Fix some stupid bug preventing cbps (callback postit) from working
655 2004-07-21 Martin Quinson
656 - Some documentation cleanups
657 - remove the useless last argument of msgtype_declare
658 - rename the Virtu functions to fit into the 'os' namespace
659 - move headers src/include -> src/include/gras/ and stop fooling with
660 gras -> . symbolic link
661 - make distcheck is now successful
663 2004-07-19 Martin Quinson
665 - Build shared library also
666 - Install html doc to the right location
667 - stop removing maintainer files in make clean
668 - build tests only on make check
670 2004-07-13 Martin Quinson
672 - No major issue in previous version => change versionning schema
673 - Re-enable little64 convertion test now that Abdou kindly regenerated the
674 corresponding dataset.
676 2004-07-11 Martin Quinson
678 - Get it working with any kind of structure (we can compute the padding
679 bytes remotely for all the architectures I have access to)
680 - Implement the structure parsing macro (still not quite robust/complete)
681 - Improvement to the remote testing toysuite
683 2004-07-10 Martin Quinson
685 - get ride of a bunch of deprecated macros
686 - actually run the test for two-compliment, not only compile it :-/
687 - test whether the structures get packed (and bail out if yes. Damn.
688 Alignment is a serious matter)
689 - test whether the structures get compacted (but respecting the alignment
690 constraints of each types)
691 - test whether the array fields of structures can straddle alignment boundaries
693 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
694 (took me 2 hours to find that bug, looking at the wrong place)
695 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
696 before sending/receiving and are used to compute the offsets of fields
697 - Given that padding size depend even on compiler options, keep track of
698 alignment and aligned_size only for the current architecture. Not a big
699 deal since we send structure fields one after the other (seems
701 - Add the datastructure used for IEEE paper by the PBIO guys to the test
702 program, let it work on linux/gcc/little32. portability todo.
704 2004-07-08 Martin Quinson
705 - import and improve remote compilation support from FAST
706 - make sure make check works on half a dozen of machines out there
708 2004-07-07 Martin Quinson
709 Let's say it's version 0.0.3 ;)
710 - Implement conversions (yuhu!)
711 - Let it work on solaris (beside conversion, of course)
712 - Stupid me, using rand() to generate the conversion datatests in not wise.
714 2004-07-06 Martin Quinson
715 - Let make dist work, since I'm gonna need it to compile on remote hosts
716 - Let Tests/datadesc_usage write the architecture on which the file was
717 generated as first byte.
718 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
719 architecture descriptions.
720 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
721 on those architectures.
722 - Optimization: send/recv array of scalar in one shoot
724 2004-07-05 Martin Quinson
725 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
727 - Plug a whole bunch of memleaks
728 - each process now have to call gras_{init,exit}. One day, their log
729 settings will be separated
730 - Continue the code factorisation between SG, RL and common in Transport.
732 2004-07-04 Martin Quinson
734 - Redistribution between SG and RL.
735 We wanna have to accept in SG, so move accepted related parts of RL in
736 the common part. (more precisely, the dynar of all known sockets is no
737 more a static in transport.c, but part of the process_data)
739 [gras_stub_generator]
740 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
742 2004-07-03 Martin Quinson
743 - Create a new log channel tbx containing dict, set, log, dynar (to shut
744 them all up in one shot)
746 - Fix the ugly case of reference to dynamic array.
747 - New (semi-public) function gras_datadesc_size to allow the messaging
748 layer to malloc the needed space for the buffer.
750 - gras_socket_close now expect the socket to close (and not its address to
751 put NULL in it after it). This is because the socket passed to handlers
752 is one of their argument (=> not writable).
754 - propagate the interface cleanup from last week in datadesc, ie remove a
755 superfluous level of indirection. User pass adress of variable
756 containing data (both when sending and receiving), and not of a variable
757 being a pointer to the data. Let's say that I like it better ;)
758 The price for that is constructs like "int msg=*(int*)payload" in
759 handlers, but it's a fine price, IMHO.
761 - Let it work in RL (yuhu)
763 2004-06-21 Martin Quinson
765 - porting SG plugin and SG select to new standards (works almost).
766 - plug memleaks and fix bugs around.
769 - cleanup the prototype of data recv and force users to specify when they
770 want to handle references to objects. Test case working even for cycles.
771 - plug memleaks. Valgrind is perfectly ok with this.
773 2004-06-12 Martin Quinson
775 - cleanup the separation between plugin and main code in plugin creation
777 2004-06-11 Martin Quinson
779 - Reput hook for raw sockets, needed for BW experiments
780 - kill a few lines of dead code
781 [Data description] Interface cleanup
782 - gras_datadesc_by_name returns the searched type or NULL.
783 That way, no variable is needed to use a type desc once, which makes
785 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
786 two parameters were strings (field name, type name), leading to
788 [Dicos] Interface cleanup
789 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
790 This is consistant with the dynar API.
792 2004-04-21 Martin Quinson
794 - Porting to new standards.
797 There is no bag anymore, no need to take extra provision to mask the
798 pointers behind "ID".
799 Better splitup of functions between files create/exchange/convert.
800 This is still a bit artificial since convert and receive are so
801 interleaved, but anyway.
803 - add a queued message list to procdata (the ones not matching criteria
805 - factorize some more code between SG and RL wrt procdata
807 - use gras_exit in example to track memleaks
808 - get rid of gs_example now that GS is properly integrated into gras
809 - update run_test to integrate the lastest tests (datadesc)
811 - rename WARNINGn macros to WARNn since it prooved error-prone
813 2004-04-19 Martin Quinson
815 - register init/exit functions within gras module mechanism
816 - send/receive function.
817 Convertion is not implemented, but short-cutted if not needed.
818 struct/array elements are sent one by one (instead of block-wise), but
819 nobody really cares (yet). Get a prototype before optimizing.
820 - tests (using a file socket) for DD send/receive on:
821 - base types: int, float
822 - array: fixed size, string (ie ref to dynamic string)
823 - structure: homogeneous, heterogeneous
824 - chained list, graph with cycle
825 Believe it or not, valgrind is not too unhappy with the results. The
826 cycle happily segfaults, but the others are ok. And I'm sick of pointers
830 - Bugfix when using a filename explicitely (instead of '-')
832 2004-04-09 Martin Quinson
834 - factorize more code between RL and SG in socket creation
835 - Complete the implementation and tests of:
837 o file (only in RL, and mainly for debugging)
839 I lost 3 days to design a portable address resolver, and then decided
840 that the prototype mainly have to run on my box.
841 Addressing portability too early may be like optimizing too early :-/
843 - use gras_init in the Tests instead of the crappy parse_log_opt
844 (the latter function is removed)
845 [Conditional execution]
846 - New functions: gras_if_RL/gras_if_SG (basic support for this)
847 [Code reorganisation]
848 - Get rid of libgrasutils.a since it makes more trouble than it solves.
849 Build examples against the RL library, since there is no way to disable
850 its creation for now.