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 communication [AL]
22 * New: MSG_task_get_source to see on which host a task was generated [HC]
24 GRAS performance improvements:
25 * Actually implement gras_datadesc_copy() so that we don't have to mimick
26 RL communication on top of SG since it's so uneffective.
27 It may also be used for inter-thread communication in RL, one day. [MQ]
28 * Use gras_datadesc_copy() to exchange messages on top of SG [MQ]
30 - improve message exchange performance on top of SG
31 - deprecate transport_plugin_sg.c:gras_trp_sg_chunk_send() & recv()
32 * Don't exchange on the network the size of the used part of buffer,
33 instead, specify the possible buffer size to read(). [MQ]
35 - reduces the amount of read/write calls (one pair per exchange)
36 - reduces the amount of exchanged data (the size)
37 - allows to retrieve all arrived data on receiver side, if we don't need
38 it right now (subsequent read will peek the buffer)
39 - allows the receiver to proceed with the begining of the stream before
41 - make it possible to build an iov transport (using readv/writev)
43 - take care of the data with non-stable storage (like stacked data),
45 * If possible, TCP send uses vector I/O (when writev() is here) [MQ]
46 - Don't use it for receive since we send data sizes and data on the
47 same stream, so we wouldn't be able to chain large amount of chunks
48 before having to flush the stuff to read the size.
49 * Rework the transport plugin mecanism to simplify it and reduce the
50 amount of pointer dereferencement when searching for the right function
53 * I guess that now, we do almost as few system calls as possible while
54 doing as few data copy as possible.
56 To improve it further, we could try to send all the sizes first and then
57 all the data (to use iov on receiving size), but it's only a partial
58 solution: when you have 2 dimensional data, the sizes of the second
59 dimension is data of the first dimension, so you need 3 streams.
61 I'm not sure the potential performance gains justify the coding burden.
65 SimGrid (3.00) stable; urgency=low
68 * New! Give the possibility to hijack the surf parser and thus bypass
69 MSG_create_environment and MSG_launch_application. Have a look at
70 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
72 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
74 SimGrid (2.96) unstable; urgency=low
79 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
80 This deprecates the xbt_error_t mecanisms.
81 It modifies (simplifies) all XBT and GRAS API.
82 MSG API keeps unchanged (exceptions raised by XBT are catched from
83 within MSG and masked with existing error handling facilities)
86 * New! Add a FATPIPE model. [AL]
87 * New! Add a parallel task model. [AL]
88 * New! Add automatically a loopback interface (in the default
89 network model) if none was precised.
92 * Bugfix: MSG_process_resume now works with the current running process.
94 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
95 * Modification of MSG_task_get_compute_duration. Once a task has been
96 processed, the value returned by this function is now equal to 0. [AL]
97 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
98 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
99 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
100 of a m_task. For now, MSG_task_cancel only works with computation
102 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
103 of the processor (in Mflop/s) regardless of the current load on the
105 * API Change: use proper naming convention for MSG_getClock and
106 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
108 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
109 This function changes the priority of a computation task. This priority
110 doesn't affect the transfer rate. A priority of 2 will make a task
111 receive two times more cpu power than the other ones. This function
112 has been added to suit the needs of Nguyen The Loc and hasn't been that
113 much tested yet. So if it fails, please report it and send me your code.
115 * API Change: removed all functions and types that were marked "deprecated"
116 since many months. Renamed MSG_global_init_args to MSG_global_init.
118 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
120 SimGrid (2.95) unstable; urgency=low
123 * Steal some nice code to GNU pth to fix context detection and usage [AL]
124 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
125 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
128 * Bugfix: Allow absolute paths for platform description files [MQ]
129 * Bugfix: do free the variables after use. Leads to drastic performance
131 * Implement max_duration (ie, timeouts) on resources [AL]
134 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
136 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
137 the needed support (timeouts) [AL]
140 * Implement measurement sockets. You can now get the bandwidth between two
141 hosts thanks to AMOK (see below). [MQ]
142 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
143 dynar over the network (yeah) [MQ]
144 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
145 * simple/static token-ring example. [Alexandre Colucci and MQ]
146 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
150 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
151 type parsing. "short" and "long" modifiers were also ignored; other
152 modifier (such as reference level) are still ignored. [MQ]
153 * Embeed the buffer size within the buffer itself on SG. [MQ]
154 That way, send() are atomic and cannot get intermixed anymore (at least
155 the ones which are less than 100k; bigger messages still have the issue)
156 * Array size pushed by the field, not by the field type (or each
157 and every long int will push stuff to the cbps) [MQ]
158 * use select() to sleep since it allows to portably sleep less than one
161 GRAS (minor cleanups)
162 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
165 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
166 * type category 'ignored' killed as it was never used and were difficult
168 * whether a type can cycle or not is now a flag, leaving room for more
169 flags (as "ignored", if we feel the need one day ;)
170 * Rename raw sockets to measurement sockets since "raw" has another
171 meaning in networking community.
174 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
175 and offers features not belonging to GRAS but that most applications
176 need. One day, it may be a set of plugins loadable at runtime.
177 * New module: bandwidth
178 bandwidth measurement between arbitrary nodes running this module. [MQ]
180 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
182 SimGrid (2.94) unstable; urgency=low
184 The first beta release of SimGrid 3 !
188 * Update the main page and the FAQ. Adding references to gforge.
191 * Add a gras_os_getpid function.
194 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
195 * Extend the logs so that they also print PID, hostname, date, ... if
197 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
198 and kill the old version which were in testsuite/
199 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
203 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
204 for users. Instead, each programming environment (GRAS, MSG,...) use it
205 when needed to provide such a feature to users.
206 Don't shortcut the mecanism or you will also shortcut the virtualization
207 you need on the simulator.
211 * Cleanups in configury with regard to compile optimization/warning flags.
212 Also add -fno-loop-optimize to any powerpc since it's the optimization
213 killing gcc (< 3.4.0).
214 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
215 needed by MSG examples complications
216 * Borrow configury beautifications from PHP
219 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
220 warning (thanks loris for stumbling into this one).
221 * Bugfix: stop loading private headers (gras_config.h) from the public
225 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
226 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
227 RL would imply a DNS resolver, which is *hard* to do in a portable way
228 (and therefore delayed).
229 * Implement a real timer mecanism and use it in timing macros. This allows
230 to avoid rounding errors and get a 0.000005 sec precision in timing
231 macros. While I was at it, various cleanups:
232 - allow to declare more than one timed section per file (fix a stupid bug)
233 - move some private declaration to the right place
234 - merge conditional execution and timing macros into emulation module
235 - document the module
236 - make sure the module cleanups its mess on gras_exit
237 * Documentation improvements:
238 - (new) how to compile applications using GRAS
239 - (new) emulation support (timing macros)
241 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
243 SimGrid (2.93) unstable; urgency=low
245 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
248 - Use Paje properly where used. Still to be sanitized properly.
249 - Portability fix: Add an implementation of the contexts using pthread
253 - Add xbt_procname(): returns the name of the current process.
254 Use it to show the current process's name in all logging.
256 - fix detection of older flex version and the reaction, since we do
257 depend on modern ones (we use lex_destroy)
258 - Better separation of SG and RL in the libs: remove all simulation code
259 from libgras. As a result, this lib is now only 200k when stripped.
260 Some of the xbt modules may also be duplicated (two sets and such) and
261 should be cleaned/killed before SG3.
262 - Insist on using xlC on AIX because of weird problems involving gcc there.
263 - Cleanup the make remote stuff. This is now done by scripts
264 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
265 mainly for our private use, but we're working on changing them to user
268 - Bugfix: flush the socket on close only if there is some *output*.
269 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
271 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
274 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
276 SimGrid (2.92) unstable; urgency=low
278 Alpha 3 on the path to SimGrid 3
282 - New! First try of benchmarking macros.
283 - New! First try so that gras_stub_generator generate deployment and
284 remote compilation helpers.
286 - Bugfix: Initialization fix in msg_test.
290 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
292 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
294 - API Change: gras_os_sleep to take the amount of seconds as a double.
295 Accepting an int was error prone since it was the only location where
296 seconds were coded as such. It leaded to damn rounding errors.
297 - Bugfix: Hard to belive that timers ever worked before this.
299 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
301 SimGrid (2.91) unstable; urgency=low
303 Alpha 2 on the path to SimGrid 3
307 - Bug fix in the lmm_solver.
309 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
310 through the function MSG_paje_output.
311 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
312 - It is possible to bound the rate of a communication in MSG with
313 MSG_task_put_bounded() (was already in the previous version but I had forgotten
314 to write it in the changelog).
315 - Bug fix to let GRAS run on top of MSG until we move it directly on top
320 - Various cleanups to the autotools stuff
321 - Begin to move Gras examples to examples/gras/
322 - Let make distcheck work again (yeah!)
324 - documentation overhauled using doxygen.
325 gtk-doc-tools is dead in SimGrid now.
326 - Automatically extract all existing logging categories, and add the list
327 to the documentation (long standing one, to say the less)
329 - Cleanup the known architecture table. Reorder the entries to group what
330 should be, and use a more consistent naming scheme.
331 (some of the test dataset are still to be regenerated)
332 - New! Allow library to register globals on each process just as userdata
334 This is implemented using a xbt_dict and not a xbt_set, so we loose the
335 lookup time (for now).
336 Use it in msg and trp.
337 This cleans a lot the internals and helps enforcing privacy of the
338 headers between the gras components.
339 - New! Add a timer mechanism, not unlike cron(8) and at(1).
340 - Bugfix: gras_os_time was delirious in RL.
341 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
343 - Reenable GRAS now that it works.
345 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
347 SimGrid (2.90) unstable; urgency=low
349 Alpha 1 on the path to SimGrid 3
351 * It is a long time since the last release of SimGrid. I'm sorry about
352 that but as I had told you, I was rewriting a lot of things. I apologize
353 to those who had been reporting bugs to me and that I had not answered.
354 If your bug is still in the new version, please tell me. Here is a
355 summary of the main changes.
357 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
358 lead by Martin Quinson. As a consequence SimGrid gains a lot in
359 portability, speed, and a lot more but you'll figure it out later.
360 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
361 I wanted to release the new MSG as soon as possible and I have
362 broken GRAS, which is the reason why, for now, only MSG is fully
363 functional. A laconic description of these projects is available
364 in the documentation.
366 * REVOLUTION 2: I have removed SG and I am now using a new simulation
367 kernel optimized for our needs (called SURF but only the developers
368 should use it). Hence, MSG is now roughly 30 times faster and I think
369 that by rewriting a little bit MSG, I could event speed it up a little
370 bit more. Beside the gain in speed, it is also much easier to encode a
371 new platform model with SURF than it was with SG. More to come...
373 * REVOLUTION 3: I have tried to change a little as possible the API of
374 MSG but a few things really had to disappear. The main differences
375 with the previous version are :
376 1) no more m_links_t and the corresponding functions. Platforms are
377 directly read from a XML description and cannot be hard-coded
378 anymore. The same format is used for application deployment
379 description. The new format is described in the documentation.
380 Have a look in tools/platform_generation. There is a tiny script
381 that converts from the old platform format to the new one. Concerning
382 the application deployment format, parsing the old one is tricky.
383 I think most of you should however be able to convert your files. If
384 it is really an issue, I can write a C code that does the conversion.
386 2) the toolbox tbx does not exist anymore. We now have a library
387 with much more data-structures but without the hash-tables (we have
388 dictionaries that are much faster).
390 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
392 *****************************************************************************
393 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
394 * don't feel like converting the oldies. (Mt) *
395 *****************************************************************************
398 Version 2.90: "the long awaited one"
399 - Finished rewriting and debugging MSG. Rewrote the documentation.
400 - disable GRAS for now since it needs to be ported to the newest SG
403 - Finish the port to windows (using mingw32 for cross-compile)
406 - Main loop and datastructures of SURF. A cpu resource object is
407 functional. Surf can thus be used to create cpu's with variable
408 performance on which you can execute some actions.
410 2004-11-15 Martin Quinson
411 - Port to ARM. Simply added the alignment and size descriptions. Should
412 work, but the ARM machines are so slow that I didn't had the opportunity
413 to 'make check' over there yet.
415 2004-11-15 Arnaud Legrand
416 - Trace manager now written. It uses a heap structure and is therefore
417 expected to be efficient. It may however be speeded up (particularly
418 when many events occur at the same date) by using red and black
419 trees. One day maybe...
420 - Max-min linear system solver written. It uses a sparse matrix
421 structure taking advantage of its expected use. Most operations are
422 O(1) and free/calloc are called as few as possible. The computation of
423 the minimum could however be improved by using a red and black tree
426 2004-11-03 Arnaud Legrand
427 - Rename every gras_* function that was in xbt/ to its xbt_
429 - Add a heap and a doubly-linked list to xbt
430 - Added a dichotomy to the dictionaries. make check works as well before
431 so I assume that the patch is correct. I do not know however if things
432 run effectively faster than before now. :)
434 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
435 SimGrid, and the version number is bumped to 2.x
437 2004-10-29 Martin Quinson
438 - Introduction of the remote errors.
439 They are the result of a RMI/RPC on the remote machine.
440 ErrCodes being scalar values, you can't get the host on which those
441 errors did happen. Extending the error mechanism as in Gnome is possible.
442 No idea yet whether it is a good idea.
444 2004-10-28 Martin Quinson
445 - Interface revolution: the Starred Structure Eradication.
446 I used to do typedef struct {} toto_t; and then handle *toto_t.
447 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
448 - ???_t is a valid type (builded with typedef)
449 - s_toto_t is a structure (access to fields with .)
450 - s_toto is a structure needing 'struct' keyword to be used
451 - e_toto_t is an enum
452 - toto_t is an 'object' (struct*)
454 typedef struct s_toto {} s_toto_t, *toto_t;
455 typedef enum {} e_toto_t;
456 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
457 s_toto_t) is private.
459 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
460 so that it can set the variable to NULL. It was so for dicts and sets,
461 it changed for dynars.
463 - Fix a bunch of memleaks in dict_remove
464 - Fix a bug in sg/server_socket opening: it failed all the time.
466 2004-10-07 Martin Quinson
467 - Speed up dynar lookup operation a bit.
469 gras_dynar_get is dead.
471 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
472 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
473 (which gives you the address of the stored data).
475 gras_dynar_get_as is an helpful macro which allows you to retrieve a
476 copy of the data using an affectation to do the job and not a memcpy.
478 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
479 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
481 It does not really speedup the dynar test because they are
482 setting elements all the time (and look them seldom). But the dict does
483 far more lookup than setting.
485 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
487 2004-10-05 Martin Quinson
488 - Allow to (en/dis)able the cycle detection at run time.
490 Whether we should check for cycle or not is now a property of each
491 datatype. When you think there may be some cycle, use datadesc_cycle_set.
492 datadesc_cycle_unset allow to remove this property when previously set.
494 Note that the cycle detection is off by default since it impacts the
495 performance. Watch the data you feed GRAS with ;)
497 This property is hereditary. Any element embedded in a structure having it
498 set have it set for the time of this data exchange.
500 You should set it both on sender and receiver side. If you don't set it on
501 sender side, it will enter an endless loop. If you forget on receiver
502 side, the cycles won't be recreated after communication.
504 - Header reorganization.
505 Kill gras_private.h, each submodule must load the headers it needs.
507 2004-10-04 Martin Quinson
508 - Interface revolution: do not try to survive to malloc failure.
510 Now, gras_malloc and friends call gras_abort() on failure.
511 As a conclusion, malloc_error is not a valid error anymore, and all
512 functions for which it was the only gras_error_t return value are
513 changed. They now return void, or there result directly.
514 This simplify the API a lot.
516 2004-09-29 Martin Quinson
517 - Re-enable raw sockets.
518 Created by gras_socket_{client,server}_ext;
519 Used with gras_raw_{send,recv}
522 It should allow to kill the last bits of gras first version soon.
524 This is not completely satisfactory yet (duplicate code with
525 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
528 - Simplify transport plugin (internal) interface by not passing any
529 argument to _server and _client, but embedding them in the socket
532 2004-09-28 Martin Quinson
533 - Finish the port to AIX.
534 autoconf was my problem (segfault within the malloc replacement
535 function. No idea why)
537 2004-09-16 Martin Quinson
538 - Fix some size_t madness on 64bit architectures.
540 2004-09-08 Martin Quinson
541 - Reduce the number of system headers loaded, overload some more system
542 calls (such as malloc to cast the result of the system one, and work
544 - Fix and reintroduce the config support
546 2004-09-07 Martin Quinson
547 - Source code reorganization to allow Arnaud to surf all over there.
548 - Allow to document the logging categories.
549 - Remove all uppercase from logging categories and useless cleanup in names.
551 2004-08-18 Martin Quinson
552 Version 0.6.2 (protocol not changed; API changed)
553 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
554 gras_error_t), and NULL when not found. Functions expecting a msgtype
555 as argument (msg_wait; msg_send) deal with NULL argument by providing a
556 hopefully usefull message.
557 - Portability to prehistoric sparcs again
559 2004-08-17 Martin Quinson
560 Version 0.6.1 (protocol not changed; ABI not changed)
561 - prealloc some buffers to speed things up
563 2004-08-11 Martin Quinson
564 Version 0.6 (protocol not changed; ABI expended)
565 - The parsing macro can deal with the references, provided that you add
566 the relevant annotations (using GRAS_ANNOTE(size,field_name))
568 2004-08-09 Martin Quinson
569 Version 0.5 (protocol not changed; ABI changed)
570 - Allow to off turn the cycle detection code in data exchange at
571 compilation time. It should be at run time, but I'm short of time (and
572 the config stuff is still broken). That way, we keep dict out of the
573 critical path, which is good because the performance is poor:
574 - search not dichotomial yet
575 - dynar give no way to access their content and memcpy everytime
576 - In composed data description (struct, ref and so on), stop foolness of
577 keeping the subtype's ID, but store the type itself. This keeps sets out
578 of the critical path, which is good since they rely on dynar and
579 dictionnaries. The only loose of that is that we cannot detect the
580 redeclaration of a structure/union with another content (but I'm not sure
581 the code detected well this error before anyway). We still can detect
582 the redefinition discrepancy for the other types.
583 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
584 it breaks the code because of type-punning used all over the place).
585 This breaks on all non-gcc architectures (for now).
587 All those changes (plus the buffer of last time) allow me to gain 2 order
588 of magnitude on cruel tests consisting of 800000 array of integers on two
589 level of a hierarchical structure (200 secondes -> 4 secondes)
592 - the selector of reference must now return the type it points to, not
595 2004-08-06 Martin Quinson
596 Version 0.4 (protocol changed; ABI not changed)
597 - Allow to pass --gras-log argument to processes in simulation mode. Really.
598 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
599 - Implement a buffer transport, and use it by default (it relies on tcp in
600 real life and on sg in simulation).
601 That's a bit hackish since I had a new field to the structure to store
602 its data without interfering with the subtype ones. Inheritance
603 is tricky in C. And that's a kind of reverse inheritance with one class
604 derivating two classes. Or maybe a game with java interfaces. Anyway,
605 that's damn hard in C (at least).
606 Moreover, I got tired while trying to ensure plugin separation and
607 genericity in SG mode. MSG wants me to do weird things, so let's go for
608 cruel hacks (temporarily of course ;).
609 See comment in transport_private.h:71
610 - do not include all the _interface headers in private but in the files
611 which really need them (to cut the compilation time when they are
614 2004-07-26 Martin Quinson
615 Version 0.3 (protocol not changed; ABI changed)
616 - Major overhault of the datadesc interface to simplify it:
617 - shorted the function names:
618 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
619 - add a trivial way to push/pop integers into the cbps without malloc.
620 This allows to make really generic sub_type description, which simply
621 pop their size of the stack.
622 - add a function gras_datadesc_ref_pop_arr() which does what users want
623 most of the time: Declare a dynamic array (which pops its size of the
624 stack) and declare a reference to it. Poor name, but anyway.
625 - kill the post-send callback, add a post-receive one
627 2004-07-23 Martin Quinson
628 Version 0.2 (protocol changed; ABI changed)
629 - add some testing for cpbs in the test cases, and fix some more bugs.
630 This invalidate again the little64 data file, since I cannot regenerate
632 - remove an awfull optimization in the logging stuff, allowing me to:
633 - understand it again
634 - learn gcc how to check that the argument match the provided format
635 - fix all errors revealed by gcc after that
636 - internal keys of dict are not \0 terminated. Deal with it properly in
637 loggings instead of segfaulting when the user want to see the logs :-/
639 2004-07-22 Martin Quinson
640 - Fix some stupid bug preventing cbps (callback postit) from working
642 2004-07-21 Martin Quinson
643 - Some documentation cleanups
644 - remove the useless last argument of msgtype_declare
645 - rename the Virtu functions to fit into the 'os' namespace
646 - move headers src/include -> src/include/gras/ and stop fooling with
647 gras -> . symbolic link
648 - make distcheck is now successful
650 2004-07-19 Martin Quinson
652 - Build shared library also
653 - Install html doc to the right location
654 - stop removing maintainer files in make clean
655 - build tests only on make check
657 2004-07-13 Martin Quinson
659 - No major issue in previous version => change versionning schema
660 - Re-enable little64 convertion test now that Abdou kindly regenerated the
661 corresponding dataset.
663 2004-07-11 Martin Quinson
665 - Get it working with any kind of structure (we can compute the padding
666 bytes remotely for all the architectures I have access to)
667 - Implement the structure parsing macro (still not quite robust/complete)
668 - Improvement to the remote testing toysuite
670 2004-07-10 Martin Quinson
672 - get ride of a bunch of deprecated macros
673 - actually run the test for two-compliment, not only compile it :-/
674 - test whether the structures get packed (and bail out if yes. Damn.
675 Alignment is a serious matter)
676 - test whether the structures get compacted (but respecting the alignment
677 constraints of each types)
678 - test whether the array fields of structures can straddle alignment boundaries
680 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
681 (took me 2 hours to find that bug, looking at the wrong place)
682 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
683 before sending/receiving and are used to compute the offsets of fields
684 - Given that padding size depend even on compiler options, keep track of
685 alignment and aligned_size only for the current architecture. Not a big
686 deal since we send structure fields one after the other (seems
688 - Add the datastructure used for IEEE paper by the PBIO guys to the test
689 program, let it work on linux/gcc/little32. portability todo.
691 2004-07-08 Martin Quinson
692 - import and improve remote compilation support from FAST
693 - make sure make check works on half a dozen of machines out there
695 2004-07-07 Martin Quinson
696 Let's say it's version 0.0.3 ;)
697 - Implement conversions (yuhu!)
698 - Let it work on solaris (beside conversion, of course)
699 - Stupid me, using rand() to generate the conversion datatests in not wise.
701 2004-07-06 Martin Quinson
702 - Let make dist work, since I'm gonna need it to compile on remote hosts
703 - Let Tests/datadesc_usage write the architecture on which the file was
704 generated as first byte.
705 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
706 architecture descriptions.
707 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
708 on those architectures.
709 - Optimization: send/recv array of scalar in one shoot
711 2004-07-05 Martin Quinson
712 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
714 - Plug a whole bunch of memleaks
715 - each process now have to call gras_{init,exit}. One day, their log
716 settings will be separated
717 - Continue the code factorisation between SG, RL and common in Transport.
719 2004-07-04 Martin Quinson
721 - Redistribution between SG and RL.
722 We wanna have to accept in SG, so move accepted related parts of RL in
723 the common part. (more precisely, the dynar of all known sockets is no
724 more a static in transport.c, but part of the process_data)
726 [gras_stub_generator]
727 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
729 2004-07-03 Martin Quinson
730 - Create a new log channel tbx containing dict, set, log, dynar (to shut
731 them all up in one shot)
733 - Fix the ugly case of reference to dynamic array.
734 - New (semi-public) function gras_datadesc_size to allow the messaging
735 layer to malloc the needed space for the buffer.
737 - gras_socket_close now expect the socket to close (and not its address to
738 put NULL in it after it). This is because the socket passed to handlers
739 is one of their argument (=> not writable).
741 - propagate the interface cleanup from last week in datadesc, ie remove a
742 superfluous level of indirection. User pass adress of variable
743 containing data (both when sending and receiving), and not of a variable
744 being a pointer to the data. Let's say that I like it better ;)
745 The price for that is constructs like "int msg=*(int*)payload" in
746 handlers, but it's a fine price, IMHO.
748 - Let it work in RL (yuhu)
750 2004-06-21 Martin Quinson
752 - porting SG plugin and SG select to new standards (works almost).
753 - plug memleaks and fix bugs around.
756 - cleanup the prototype of data recv and force users to specify when they
757 want to handle references to objects. Test case working even for cycles.
758 - plug memleaks. Valgrind is perfectly ok with this.
760 2004-06-12 Martin Quinson
762 - cleanup the separation between plugin and main code in plugin creation
764 2004-06-11 Martin Quinson
766 - Reput hook for raw sockets, needed for BW experiments
767 - kill a few lines of dead code
768 [Data description] Interface cleanup
769 - gras_datadesc_by_name returns the searched type or NULL.
770 That way, no variable is needed to use a type desc once, which makes
772 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
773 two parameters were strings (field name, type name), leading to
775 [Dicos] Interface cleanup
776 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
777 This is consistant with the dynar API.
779 2004-04-21 Martin Quinson
781 - Porting to new standards.
784 There is no bag anymore, no need to take extra provision to mask the
785 pointers behind "ID".
786 Better splitup of functions between files create/exchange/convert.
787 This is still a bit artificial since convert and receive are so
788 interleaved, but anyway.
790 - add a queued message list to procdata (the ones not matching criteria
792 - factorize some more code between SG and RL wrt procdata
794 - use gras_exit in example to track memleaks
795 - get rid of gs_example now that GS is properly integrated into gras
796 - update run_test to integrate the lastest tests (datadesc)
798 - rename WARNINGn macros to WARNn since it prooved error-prone
800 2004-04-19 Martin Quinson
802 - register init/exit functions within gras module mechanism
803 - send/receive function.
804 Convertion is not implemented, but short-cutted if not needed.
805 struct/array elements are sent one by one (instead of block-wise), but
806 nobody really cares (yet). Get a prototype before optimizing.
807 - tests (using a file socket) for DD send/receive on:
808 - base types: int, float
809 - array: fixed size, string (ie ref to dynamic string)
810 - structure: homogeneous, heterogeneous
811 - chained list, graph with cycle
812 Believe it or not, valgrind is not too unhappy with the results. The
813 cycle happily segfaults, but the others are ok. And I'm sick of pointers
817 - Bugfix when using a filename explicitely (instead of '-')
819 2004-04-09 Martin Quinson
821 - factorize more code between RL and SG in socket creation
822 - Complete the implementation and tests of:
824 o file (only in RL, and mainly for debugging)
826 I lost 3 days to design a portable address resolver, and then decided
827 that the prototype mainly have to run on my box.
828 Addressing portability too early may be like optimizing too early :-/
830 - use gras_init in the Tests instead of the crappy parse_log_opt
831 (the latter function is removed)
832 [Conditional execution]
833 - New functions: gras_if_RL/gras_if_SG (basic support for this)
834 [Code reorganisation]
835 - Get rid of libgrasutils.a since it makes more trouble than it solves.
836 Build examples against the RL library, since there is no way to disable
837 its creation for now.