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]
21 * Add addtionnal checkings on channel values in communicating functions.
23 GRAS performance improvements:
24 * Actually implement gras_datadesc_copy() so that we don't have to mimick
25 RL communication on top of SG since it's so uneffective.
26 It may also be used for inter-thread communication in RL, one day. [MQ]
27 * Use gras_datadesc_copy() to exchange messages on top of SG [MQ]
29 - improve message exchange performance on top of SG
30 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
31 * Don't exchange on the network the size of the used part of buffer,
32 instead, specify the possible buffer size to read(). [MQ]
34 - reduces the amount of read/write calls (one pair per exchange)
35 - reduces the amount of exchanged data (the size)
36 - allows to retrieve all arrived data on receiver side, if we don't need
37 it right now (subsequent read will peek the buffer)
38 - allows the receiver to proceed with the begining of the stream before
40 - make it possible to build an iov transport (using readv/writev)
42 - take care of the data with non-stable storage (like stacked data),
44 * If possible, TCP send uses vector I/O (when writev() is here) [MQ]
45 - Don't use it for receive since we send data sizes and data on the
46 same stream, so we wouldn't be able to chain large amount of chunks
47 before having to flush the stuff to read the size.
48 * Rework the transport plugin mecanism to simplify it and reduce the
49 amount of pointer dereferencement when searching for the right function
52 * I guess that now, we do almost as few system calls as possible while
53 doing as few data copy as possible.
55 To improve it further, we could try to send all the sizes first and then
56 all the data (to use iov on receiving size), but it's only a partial
57 solution: when you have 2 dimensional data, the sizes of the second
58 dimension is data of the first dimension, so you need 3 streams.
60 I'm not sure the potential performance gains justify the coding burden.
64 SimGrid (3.00) stable; urgency=low
67 * New! Give the possibility to hijack the surf parser and thus bypass
68 MSG_create_environment and MSG_launch_application. Have a look at
69 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
71 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
73 SimGrid (2.96) unstable; urgency=low
78 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
79 This deprecates the xbt_error_t mecanisms.
80 It modifies (simplifies) all XBT and GRAS API.
81 MSG API keeps unchanged (exceptions raised by XBT are catched from
82 within MSG and masked with existing error handling facilities)
85 * New! Add a FATPIPE model. [AL]
86 * New! Add a parallel task model. [AL]
87 * New! Add automatically a loopback interface (in the default
88 network model) if none was precised.
91 * Bugfix: MSG_process_resume now works with the current running process.
93 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
94 * Modification of MSG_task_get_compute_duration. Once a task has been
95 processed, the value returned by this function is now equal to 0. [AL]
96 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
97 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
98 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
99 of a m_task. For now, MSG_task_cancel only works with computation
101 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
102 of the processor (in Mflop/s) regardless of the current load on the
104 * API Change: use proper naming convention for MSG_getClock and
105 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
107 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
108 This function changes the priority of a computation task. This priority
109 doesn't affect the transfer rate. A priority of 2 will make a task
110 receive two times more cpu power than the other ones. This function
111 has been added to suit the needs of Nguyen The Loc and hasn't been that
112 much tested yet. So if it fails, please report it and send me your code.
114 * API Change: removed all functions and types that were marked "deprecated"
115 since many months. Renamed MSG_global_init_args to MSG_global_init.
117 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
119 SimGrid (2.95) unstable; urgency=low
122 * Steal some nice code to GNU pth to fix context detection and usage [AL]
123 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
124 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
127 * Bugfix: Allow absolute paths for platform description files [MQ]
128 * Bugfix: do free the variables after use. Leads to drastic performance
130 * Implement max_duration (ie, timeouts) on resources [AL]
133 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
135 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
136 the needed support (timeouts) [AL]
139 * Implement measurement sockets. You can now get the bandwidth between two
140 hosts thanks to AMOK (see below). [MQ]
141 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
142 dynar over the network (yeah) [MQ]
143 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
144 * simple/static token-ring example. [Alexandre Colucci and MQ]
145 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
149 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
150 type parsing. "short" and "long" modifiers were also ignored; other
151 modifier (such as reference level) are still ignored. [MQ]
152 * Embeed the buffer size within the buffer itself on SG. [MQ]
153 That way, send() are atomic and cannot get intermixed anymore (at least
154 the ones which are less than 100k; bigger messages still have the issue)
155 * Array size pushed by the field, not by the field type (or each
156 and every long int will push stuff to the cbps) [MQ]
157 * use select() to sleep since it allows to portably sleep less than one
160 GRAS (minor cleanups)
161 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
164 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
165 * type category 'ignored' killed as it was never used and were difficult
167 * whether a type can cycle or not is now a flag, leaving room for more
168 flags (as "ignored", if we feel the need one day ;)
169 * Rename raw sockets to measurement sockets since "raw" has another
170 meaning in networking community.
173 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
174 and offers features not belonging to GRAS but that most applications
175 need. One day, it may be a set of plugins loadable at runtime.
176 * New module: bandwidth
177 bandwidth measurement between arbitrary nodes running this module. [MQ]
179 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
181 SimGrid (2.94) unstable; urgency=low
183 The first beta release of SimGrid 3 !
187 * Update the main page and the FAQ. Adding references to gforge.
190 * Add a gras_os_getpid function.
193 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
194 * Extend the logs so that they also print PID, hostname, date, ... if
196 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
197 and kill the old version which were in testsuite/
198 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
202 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
203 for users. Instead, each programming environment (GRAS, MSG,...) use it
204 when needed to provide such a feature to users.
205 Don't shortcut the mecanism or you will also shortcut the virtualization
206 you need on the simulator.
210 * Cleanups in configury with regard to compile optimization/warning flags.
211 Also add -fno-loop-optimize to any powerpc since it's the optimization
212 killing gcc (< 3.4.0).
213 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
214 needed by MSG examples complications
215 * Borrow configury beautifications from PHP
218 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
219 warning (thanks loris for stumbling into this one).
220 * Bugfix: stop loading private headers (gras_config.h) from the public
224 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
225 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
226 RL would imply a DNS resolver, which is *hard* to do in a portable way
227 (and therefore delayed).
228 * Implement a real timer mecanism and use it in timing macros. This allows
229 to avoid rounding errors and get a 0.000005 sec precision in timing
230 macros. While I was at it, various cleanups:
231 - allow to declare more than one timed section per file (fix a stupid bug)
232 - move some private declaration to the right place
233 - merge conditional execution and timing macros into emulation module
234 - document the module
235 - make sure the module cleanups its mess on gras_exit
236 * Documentation improvements:
237 - (new) how to compile applications using GRAS
238 - (new) emulation support (timing macros)
240 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
242 SimGrid (2.93) unstable; urgency=low
244 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
247 - Use Paje properly where used. Still to be sanitized properly.
248 - Portability fix: Add an implementation of the contexts using pthread
252 - Add xbt_procname(): returns the name of the current process.
253 Use it to show the current process's name in all logging.
255 - fix detection of older flex version and the reaction, since we do
256 depend on modern ones (we use lex_destroy)
257 - Better separation of SG and RL in the libs: remove all simulation code
258 from libgras. As a result, this lib is now only 200k when stripped.
259 Some of the xbt modules may also be duplicated (two sets and such) and
260 should be cleaned/killed before SG3.
261 - Insist on using xlC on AIX because of weird problems involving gcc there.
262 - Cleanup the make remote stuff. This is now done by scripts
263 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
264 mainly for our private use, but we're working on changing them to user
267 - Bugfix: flush the socket on close only if there is some *output*.
268 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
270 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
273 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
275 SimGrid (2.92) unstable; urgency=low
277 Alpha 3 on the path to SimGrid 3
281 - New! First try of benchmarking macros.
282 - New! First try so that gras_stub_generator generate deployment and
283 remote compilation helpers.
285 - Bugfix: Initialization fix in msg_test.
289 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
291 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
293 - API Change: gras_os_sleep to take the amount of seconds as a double.
294 Accepting an int was error prone since it was the only location where
295 seconds were coded as such. It leaded to damn rounding errors.
296 - Bugfix: Hard to belive that timers ever worked before this.
298 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
300 SimGrid (2.91) unstable; urgency=low
302 Alpha 2 on the path to SimGrid 3
306 - Bug fix in the lmm_solver.
308 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
309 through the function MSG_paje_output.
310 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
311 - It is possible to bound the rate of a communication in MSG with
312 MSG_task_put_bounded() (was already in the previous version but I had forgotten
313 to write it in the changelog).
314 - Bug fix to let GRAS run on top of MSG until we move it directly on top
319 - Various cleanups to the autotools stuff
320 - Begin to move Gras examples to examples/gras/
321 - Let make distcheck work again (yeah!)
323 - documentation overhauled using doxygen.
324 gtk-doc-tools is dead in SimGrid now.
325 - Automatically extract all existing logging categories, and add the list
326 to the documentation (long standing one, to say the less)
328 - Cleanup the known architecture table. Reorder the entries to group what
329 should be, and use a more consistent naming scheme.
330 (some of the test dataset are still to be regenerated)
331 - New! Allow library to register globals on each process just as userdata
333 This is implemented using a xbt_dict and not a xbt_set, so we loose the
334 lookup time (for now).
335 Use it in msg and trp.
336 This cleans a lot the internals and helps enforcing privacy of the
337 headers between the gras components.
338 - New! Add a timer mechanism, not unlike cron(8) and at(1).
339 - Bugfix: gras_os_time was delirious in RL.
340 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
342 - Reenable GRAS now that it works.
344 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
346 SimGrid (2.90) unstable; urgency=low
348 Alpha 1 on the path to SimGrid 3
350 * It is a long time since the last release of SimGrid. I'm sorry about
351 that but as I had told you, I was rewriting a lot of things. I apologize
352 to those who had been reporting bugs to me and that I had not answered.
353 If your bug is still in the new version, please tell me. Here is a
354 summary of the main changes.
356 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
357 lead by Martin Quinson. As a consequence SimGrid gains a lot in
358 portability, speed, and a lot more but you'll figure it out later.
359 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
360 I wanted to release the new MSG as soon as possible and I have
361 broken GRAS, which is the reason why, for now, only MSG is fully
362 functional. A laconic description of these projects is available
363 in the documentation.
365 * REVOLUTION 2: I have removed SG and I am now using a new simulation
366 kernel optimized for our needs (called SURF but only the developers
367 should use it). Hence, MSG is now roughly 30 times faster and I think
368 that by rewriting a little bit MSG, I could event speed it up a little
369 bit more. Beside the gain in speed, it is also much easier to encode a
370 new platform model with SURF than it was with SG. More to come...
372 * REVOLUTION 3: I have tried to change a little as possible the API of
373 MSG but a few things really had to disappear. The main differences
374 with the previous version are :
375 1) no more m_links_t and the corresponding functions. Platforms are
376 directly read from a XML description and cannot be hard-coded
377 anymore. The same format is used for application deployment
378 description. The new format is described in the documentation.
379 Have a look in tools/platform_generation. There is a tiny script
380 that converts from the old platform format to the new one. Concerning
381 the application deployment format, parsing the old one is tricky.
382 I think most of you should however be able to convert your files. If
383 it is really an issue, I can write a C code that does the conversion.
385 2) the toolbox tbx does not exist anymore. We now have a library
386 with much more data-structures but without the hash-tables (we have
387 dictionaries that are much faster).
389 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
391 *****************************************************************************
392 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
393 * don't feel like converting the oldies. (Mt) *
394 *****************************************************************************
397 Version 2.90: "the long awaited one"
398 - Finished rewriting and debugging MSG. Rewrote the documentation.
399 - disable GRAS for now since it needs to be ported to the newest SG
402 - Finish the port to windows (using mingw32 for cross-compile)
405 - Main loop and datastructures of SURF. A cpu resource object is
406 functional. Surf can thus be used to create cpu's with variable
407 performance on which you can execute some actions.
409 2004-11-15 Martin Quinson
410 - Port to ARM. Simply added the alignment and size descriptions. Should
411 work, but the ARM machines are so slow that I didn't had the opportunity
412 to 'make check' over there yet.
414 2004-11-15 Arnaud Legrand
415 - Trace manager now written. It uses a heap structure and is therefore
416 expected to be efficient. It may however be speeded up (particularly
417 when many events occur at the same date) by using red and black
418 trees. One day maybe...
419 - Max-min linear system solver written. It uses a sparse matrix
420 structure taking advantage of its expected use. Most operations are
421 O(1) and free/calloc are called as few as possible. The computation of
422 the minimum could however be improved by using a red and black tree
425 2004-11-03 Arnaud Legrand
426 - Rename every gras_* function that was in xbt/ to its xbt_
428 - Add a heap and a doubly-linked list to xbt
429 - Added a dichotomy to the dictionaries. make check works as well before
430 so I assume that the patch is correct. I do not know however if things
431 run effectively faster than before now. :)
433 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
434 SimGrid, and the version number is bumped to 2.x
436 2004-10-29 Martin Quinson
437 - Introduction of the remote errors.
438 They are the result of a RMI/RPC on the remote machine.
439 ErrCodes being scalar values, you can't get the host on which those
440 errors did happen. Extending the error mechanism as in Gnome is possible.
441 No idea yet whether it is a good idea.
443 2004-10-28 Martin Quinson
444 - Interface revolution: the Starred Structure Eradication.
445 I used to do typedef struct {} toto_t; and then handle *toto_t.
446 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
447 - ???_t is a valid type (builded with typedef)
448 - s_toto_t is a structure (access to fields with .)
449 - s_toto is a structure needing 'struct' keyword to be used
450 - e_toto_t is an enum
451 - toto_t is an 'object' (struct*)
453 typedef struct s_toto {} s_toto_t, *toto_t;
454 typedef enum {} e_toto_t;
455 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
456 s_toto_t) is private.
458 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
459 so that it can set the variable to NULL. It was so for dicts and sets,
460 it changed for dynars.
462 - Fix a bunch of memleaks in dict_remove
463 - Fix a bug in sg/server_socket opening: it failed all the time.
465 2004-10-07 Martin Quinson
466 - Speed up dynar lookup operation a bit.
468 gras_dynar_get is dead.
470 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
471 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
472 (which gives you the address of the stored data).
474 gras_dynar_get_as is an helpful macro which allows you to retrieve a
475 copy of the data using an affectation to do the job and not a memcpy.
477 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
478 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
480 It does not really speedup the dynar test because they are
481 setting elements all the time (and look them seldom). But the dict does
482 far more lookup than setting.
484 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
486 2004-10-05 Martin Quinson
487 - Allow to (en/dis)able the cycle detection at run time.
489 Whether we should check for cycle or not is now a property of each
490 datatype. When you think there may be some cycle, use datadesc_cycle_set.
491 datadesc_cycle_unset allow to remove this property when previously set.
493 Note that the cycle detection is off by default since it impacts the
494 performance. Watch the data you feed GRAS with ;)
496 This property is hereditary. Any element embedded in a structure having it
497 set have it set for the time of this data exchange.
499 You should set it both on sender and receiver side. If you don't set it on
500 sender side, it will enter an endless loop. If you forget on receiver
501 side, the cycles won't be recreated after communication.
503 - Header reorganization.
504 Kill gras_private.h, each submodule must load the headers it needs.
506 2004-10-04 Martin Quinson
507 - Interface revolution: do not try to survive to malloc failure.
509 Now, gras_malloc and friends call gras_abort() on failure.
510 As a conclusion, malloc_error is not a valid error anymore, and all
511 functions for which it was the only gras_error_t return value are
512 changed. They now return void, or there result directly.
513 This simplify the API a lot.
515 2004-09-29 Martin Quinson
516 - Re-enable raw sockets.
517 Created by gras_socket_{client,server}_ext;
518 Used with gras_raw_{send,recv}
521 It should allow to kill the last bits of gras first version soon.
523 This is not completely satisfactory yet (duplicate code with
524 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
527 - Simplify transport plugin (internal) interface by not passing any
528 argument to _server and _client, but embedding them in the socket
531 2004-09-28 Martin Quinson
532 - Finish the port to AIX.
533 autoconf was my problem (segfault within the malloc replacement
534 function. No idea why)
536 2004-09-16 Martin Quinson
537 - Fix some size_t madness on 64bit architectures.
539 2004-09-08 Martin Quinson
540 - Reduce the number of system headers loaded, overload some more system
541 calls (such as malloc to cast the result of the system one, and work
543 - Fix and reintroduce the config support
545 2004-09-07 Martin Quinson
546 - Source code reorganization to allow Arnaud to surf all over there.
547 - Allow to document the logging categories.
548 - Remove all uppercase from logging categories and useless cleanup in names.
550 2004-08-18 Martin Quinson
551 Version 0.6.2 (protocol not changed; API changed)
552 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
553 gras_error_t), and NULL when not found. Functions expecting a msgtype
554 as argument (msg_wait; msg_send) deal with NULL argument by providing a
555 hopefully usefull message.
556 - Portability to prehistoric sparcs again
558 2004-08-17 Martin Quinson
559 Version 0.6.1 (protocol not changed; ABI not changed)
560 - prealloc some buffers to speed things up
562 2004-08-11 Martin Quinson
563 Version 0.6 (protocol not changed; ABI expended)
564 - The parsing macro can deal with the references, provided that you add
565 the relevant annotations (using GRAS_ANNOTE(size,field_name))
567 2004-08-09 Martin Quinson
568 Version 0.5 (protocol not changed; ABI changed)
569 - Allow to off turn the cycle detection code in data exchange at
570 compilation time. It should be at run time, but I'm short of time (and
571 the config stuff is still broken). That way, we keep dict out of the
572 critical path, which is good because the performance is poor:
573 - search not dichotomial yet
574 - dynar give no way to access their content and memcpy everytime
575 - In composed data description (struct, ref and so on), stop foolness of
576 keeping the subtype's ID, but store the type itself. This keeps sets out
577 of the critical path, which is good since they rely on dynar and
578 dictionnaries. The only loose of that is that we cannot detect the
579 redeclaration of a structure/union with another content (but I'm not sure
580 the code detected well this error before anyway). We still can detect
581 the redefinition discrepancy for the other types.
582 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
583 it breaks the code because of type-punning used all over the place).
584 This breaks on all non-gcc architectures (for now).
586 All those changes (plus the buffer of last time) allow me to gain 2 order
587 of magnitude on cruel tests consisting of 800000 array of integers on two
588 level of a hierarchical structure (200 secondes -> 4 secondes)
591 - the selector of reference must now return the type it points to, not
594 2004-08-06 Martin Quinson
595 Version 0.4 (protocol changed; ABI not changed)
596 - Allow to pass --gras-log argument to processes in simulation mode. Really.
597 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
598 - Implement a buffer transport, and use it by default (it relies on tcp in
599 real life and on sg in simulation).
600 That's a bit hackish since I had a new field to the structure to store
601 its data without interfering with the subtype ones. Inheritance
602 is tricky in C. And that's a kind of reverse inheritance with one class
603 derivating two classes. Or maybe a game with java interfaces. Anyway,
604 that's damn hard in C (at least).
605 Moreover, I got tired while trying to ensure plugin separation and
606 genericity in SG mode. MSG wants me to do weird things, so let's go for
607 cruel hacks (temporarily of course ;).
608 See comment in transport_private.h:71
609 - do not include all the _interface headers in private but in the files
610 which really need them (to cut the compilation time when they are
613 2004-07-26 Martin Quinson
614 Version 0.3 (protocol not changed; ABI changed)
615 - Major overhault of the datadesc interface to simplify it:
616 - shorted the function names:
617 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
618 - add a trivial way to push/pop integers into the cbps without malloc.
619 This allows to make really generic sub_type description, which simply
620 pop their size of the stack.
621 - add a function gras_datadesc_ref_pop_arr() which does what users want
622 most of the time: Declare a dynamic array (which pops its size of the
623 stack) and declare a reference to it. Poor name, but anyway.
624 - kill the post-send callback, add a post-receive one
626 2004-07-23 Martin Quinson
627 Version 0.2 (protocol changed; ABI changed)
628 - add some testing for cpbs in the test cases, and fix some more bugs.
629 This invalidate again the little64 data file, since I cannot regenerate
631 - remove an awfull optimization in the logging stuff, allowing me to:
632 - understand it again
633 - learn gcc how to check that the argument match the provided format
634 - fix all errors revealed by gcc after that
635 - internal keys of dict are not \0 terminated. Deal with it properly in
636 loggings instead of segfaulting when the user want to see the logs :-/
638 2004-07-22 Martin Quinson
639 - Fix some stupid bug preventing cbps (callback postit) from working
641 2004-07-21 Martin Quinson
642 - Some documentation cleanups
643 - remove the useless last argument of msgtype_declare
644 - rename the Virtu functions to fit into the 'os' namespace
645 - move headers src/include -> src/include/gras/ and stop fooling with
646 gras -> . symbolic link
647 - make distcheck is now successful
649 2004-07-19 Martin Quinson
651 - Build shared library also
652 - Install html doc to the right location
653 - stop removing maintainer files in make clean
654 - build tests only on make check
656 2004-07-13 Martin Quinson
658 - No major issue in previous version => change versionning schema
659 - Re-enable little64 convertion test now that Abdou kindly regenerated the
660 corresponding dataset.
662 2004-07-11 Martin Quinson
664 - Get it working with any kind of structure (we can compute the padding
665 bytes remotely for all the architectures I have access to)
666 - Implement the structure parsing macro (still not quite robust/complete)
667 - Improvement to the remote testing toysuite
669 2004-07-10 Martin Quinson
671 - get ride of a bunch of deprecated macros
672 - actually run the test for two-compliment, not only compile it :-/
673 - test whether the structures get packed (and bail out if yes. Damn.
674 Alignment is a serious matter)
675 - test whether the structures get compacted (but respecting the alignment
676 constraints of each types)
677 - test whether the array fields of structures can straddle alignment boundaries
679 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
680 (took me 2 hours to find that bug, looking at the wrong place)
681 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
682 before sending/receiving and are used to compute the offsets of fields
683 - Given that padding size depend even on compiler options, keep track of
684 alignment and aligned_size only for the current architecture. Not a big
685 deal since we send structure fields one after the other (seems
687 - Add the datastructure used for IEEE paper by the PBIO guys to the test
688 program, let it work on linux/gcc/little32. portability todo.
690 2004-07-08 Martin Quinson
691 - import and improve remote compilation support from FAST
692 - make sure make check works on half a dozen of machines out there
694 2004-07-07 Martin Quinson
695 Let's say it's version 0.0.3 ;)
696 - Implement conversions (yuhu!)
697 - Let it work on solaris (beside conversion, of course)
698 - Stupid me, using rand() to generate the conversion datatests in not wise.
700 2004-07-06 Martin Quinson
701 - Let make dist work, since I'm gonna need it to compile on remote hosts
702 - Let Tests/datadesc_usage write the architecture on which the file was
703 generated as first byte.
704 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
705 architecture descriptions.
706 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
707 on those architectures.
708 - Optimization: send/recv array of scalar in one shoot
710 2004-07-05 Martin Quinson
711 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
713 - Plug a whole bunch of memleaks
714 - each process now have to call gras_{init,exit}. One day, their log
715 settings will be separated
716 - Continue the code factorisation between SG, RL and common in Transport.
718 2004-07-04 Martin Quinson
720 - Redistribution between SG and RL.
721 We wanna have to accept in SG, so move accepted related parts of RL in
722 the common part. (more precisely, the dynar of all known sockets is no
723 more a static in transport.c, but part of the process_data)
725 [gras_stub_generator]
726 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
728 2004-07-03 Martin Quinson
729 - Create a new log channel tbx containing dict, set, log, dynar (to shut
730 them all up in one shot)
732 - Fix the ugly case of reference to dynamic array.
733 - New (semi-public) function gras_datadesc_size to allow the messaging
734 layer to malloc the needed space for the buffer.
736 - gras_socket_close now expect the socket to close (and not its address to
737 put NULL in it after it). This is because the socket passed to handlers
738 is one of their argument (=> not writable).
740 - propagate the interface cleanup from last week in datadesc, ie remove a
741 superfluous level of indirection. User pass adress of variable
742 containing data (both when sending and receiving), and not of a variable
743 being a pointer to the data. Let's say that I like it better ;)
744 The price for that is constructs like "int msg=*(int*)payload" in
745 handlers, but it's a fine price, IMHO.
747 - Let it work in RL (yuhu)
749 2004-06-21 Martin Quinson
751 - porting SG plugin and SG select to new standards (works almost).
752 - plug memleaks and fix bugs around.
755 - cleanup the prototype of data recv and force users to specify when they
756 want to handle references to objects. Test case working even for cycles.
757 - plug memleaks. Valgrind is perfectly ok with this.
759 2004-06-12 Martin Quinson
761 - cleanup the separation between plugin and main code in plugin creation
763 2004-06-11 Martin Quinson
765 - Reput hook for raw sockets, needed for BW experiments
766 - kill a few lines of dead code
767 [Data description] Interface cleanup
768 - gras_datadesc_by_name returns the searched type or NULL.
769 That way, no variable is needed to use a type desc once, which makes
771 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
772 two parameters were strings (field name, type name), leading to
774 [Dicos] Interface cleanup
775 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
776 This is consistant with the dynar API.
778 2004-04-21 Martin Quinson
780 - Porting to new standards.
783 There is no bag anymore, no need to take extra provision to mask the
784 pointers behind "ID".
785 Better splitup of functions between files create/exchange/convert.
786 This is still a bit artificial since convert and receive are so
787 interleaved, but anyway.
789 - add a queued message list to procdata (the ones not matching criteria
791 - factorize some more code between SG and RL wrt procdata
793 - use gras_exit in example to track memleaks
794 - get rid of gs_example now that GS is properly integrated into gras
795 - update run_test to integrate the lastest tests (datadesc)
797 - rename WARNINGn macros to WARNn since it prooved error-prone
799 2004-04-19 Martin Quinson
801 - register init/exit functions within gras module mechanism
802 - send/receive function.
803 Convertion is not implemented, but short-cutted if not needed.
804 struct/array elements are sent one by one (instead of block-wise), but
805 nobody really cares (yet). Get a prototype before optimizing.
806 - tests (using a file socket) for DD send/receive on:
807 - base types: int, float
808 - array: fixed size, string (ie ref to dynamic string)
809 - structure: homogeneous, heterogeneous
810 - chained list, graph with cycle
811 Believe it or not, valgrind is not too unhappy with the results. The
812 cycle happily segfaults, but the others are ok. And I'm sick of pointers
816 - Bugfix when using a filename explicitely (instead of '-')
818 2004-04-09 Martin Quinson
820 - factorize more code between RL and SG in socket creation
821 - Complete the implementation and tests of:
823 o file (only in RL, and mainly for debugging)
825 I lost 3 days to design a portable address resolver, and then decided
826 that the prototype mainly have to run on my box.
827 Addressing portability too early may be like optimizing too early :-/
829 - use gras_init in the Tests instead of the crappy parse_log_opt
830 (the latter function is removed)
831 [Conditional execution]
832 - New functions: gras_if_RL/gras_if_SG (basic support for this)
833 [Code reorganisation]
834 - Get rid of libgrasutils.a since it makes more trouble than it solves.
835 Build examples against the RL library, since there is no way to disable
836 its creation for now.