1 SimGrid (2.97) unstable; urgency=low
6 SimGrid (2.96) unstable; urgency=low
11 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
12 This deprecates the xbt_error_t mecanisms.
13 It modifies (simplifies) all XBT and GRAS API.
14 MSG API keeps unchanged (exceptions raised by XBT are catched from
15 within MSG and masked with existing error handling facilities)
18 * New! Add a FATPIPE model. [AL]
19 * New! Add a parallel task model. [AL]
20 * New! Add automatically a loopback interface (in the default
21 network model) if none was precised.
24 * Bugfix: MSG_process_resume now works with the current running process.
26 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
27 * Modification of MSG_task_get_compute_duration. Once a task has been
28 processed, the value returned by this function is now equal to 0. [AL]
29 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
30 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
31 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
32 of a m_task. For now, MSG_task_cancel only works with computation
34 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
35 of the processor (in Mflop/s) regardless of the current load on the
37 * API Change: use proper naming convention for MSG_getClock and
38 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
40 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
41 This function changes the priority of a computation task. This priority
42 doesn't affect the transfer rate. A priority of 2 will make a task
43 receive two times more cpu power than the other ones. This function
44 has been added to suit the needs of Nguyen The Loc and hasn't been that
45 much tested yet. So if it fails, please report it and send me your code.
47 * API Change: removed all functions and types that were marked "deprecated"
48 since many months. Renamed MSG_global_init_args to MSG_global_init.
50 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
52 SimGrid (2.95) unstable; urgency=low
55 * Steal some nice code to GNU pth to fix context detection and usage [AL]
56 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
57 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
60 * Bugfix: Allow absolute paths for platform description files [MQ]
61 * Bugfix: do free the variables after use. Leads to drastic performance
63 * Implement max_duration (ie, timeouts) on resources [AL]
66 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
68 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
69 the needed support (timeouts) [AL]
72 * Implement measurement sockets. You can now get the bandwidth between two
73 hosts thanks to AMOK (see below). [MQ]
74 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
75 dynar over the network (yeah) [MQ]
76 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
77 * simple/static token-ring example. [Alexandre Colucci and MQ]
78 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
82 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
83 type parsing. "short" and "long" modifiers were also ignored; other
84 modifier (such as reference level) are still ignored. [MQ]
85 * Embeed the buffer size within the buffer itself on SG. [MQ]
86 That way, send() are atomic and cannot get intermixed anymore (at least
87 the ones which are less than 100k; bigger messages still have the issue)
88 * Array size pushed by the field, not by the field type (or each
89 and every long int will push stuff to the cbps) [MQ]
90 * use select() to sleep since it allows to portably sleep less than one
94 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
97 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
98 * type category 'ignored' killed as it was never used and were difficult
100 * whether a type can cycle or not is now a flag, leaving room for more
101 flags (as "ignored", if we feel the need one day ;)
102 * Rename raw sockets to measurement sockets since "raw" has another
103 meaning in networking community.
106 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
107 and offers features not belonging to GRAS but that most applications
108 need. One day, it may be a set of plugins loadable at runtime.
109 * New module: bandwidth
110 bandwidth measurement between arbitrary nodes running this module. [MQ]
112 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
114 SimGrid (2.94) unstable; urgency=low
116 The first beta release of SimGrid 3 !
120 * Update the main page and the FAQ. Adding references to gforge.
123 * Add a gras_os_getpid function.
126 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
127 * Extend the logs so that they also print PID, hostname, date, ... if
129 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
130 and kill the old version which were in testsuite/
131 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
135 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
136 for users. Instead, each programming environment (GRAS, MSG,...) use it
137 when needed to provide such a feature to users.
138 Don't shortcut the mecanism or you will also shortcut the virtualization
139 you need on the simulator.
143 * Cleanups in configury with regard to compile optimization/warning flags.
144 Also add -fno-loop-optimize to any powerpc since it's the optimization
145 killing gcc (< 3.4.0).
146 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
147 needed by MSG examples complications
148 * Borrow configury beautifications from PHP
151 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
152 warning (thanks loris for stumbling into this one).
153 * Bugfix: stop loading private headers (gras_config.h) from the public
157 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
158 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
159 RL would imply a DNS resolver, which is *hard* to do in a portable way
160 (and therefore delayed).
161 * Implement a real timer mecanism and use it in timing macros. This allows
162 to avoid rounding errors and get a 0.000005 sec precision in timing
163 macros. While I was at it, various cleanups:
164 - allow to declare more than one timed section per file (fix a stupid bug)
165 - move some private declaration to the right place
166 - merge conditional execution and timing macros into emulation module
167 - document the module
168 - make sure the module cleanups its mess on gras_exit
169 * Documentation improvements:
170 - (new) how to compile applications using GRAS
171 - (new) emulation support (timing macros)
173 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
175 SimGrid (2.93) unstable; urgency=low
177 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
180 - Use Paje properly where used. Still to be sanitized properly.
181 - Portability fix: Add an implementation of the contexts using pthread
185 - Add xbt_procname(): returns the name of the current process.
186 Use it to show the current process's name in all logging.
188 - fix detection of older flex version and the reaction, since we do
189 depend on modern ones (we use lex_destroy)
190 - Better separation of SG and RL in the libs: remove all simulation code
191 from libgras. As a result, this lib is now only 200k when stripped.
192 Some of the xbt modules may also be duplicated (two sets and such) and
193 should be cleaned/killed before SG3.
194 - Insist on using xlC on AIX because of weird problems involving gcc there.
195 - Cleanup the make remote stuff. This is now done by scripts
196 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
197 mainly for our private use, but we're working on changing them to user
200 - Bugfix: flush the socket on close only if there is some *output*.
201 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
203 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
206 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
208 SimGrid (2.92) unstable; urgency=low
210 Alpha 3 on the path to SimGrid 3
214 - New! First try of benchmarking macros.
215 - New! First try so that gras_stub_generator generate deployment and
216 remote compilation helpers.
218 - Bugfix: Initialization fix in msg_test.
222 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
224 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
226 - API Change: gras_os_sleep to take the amount of seconds as a double.
227 Accepting an int was error prone since it was the only location where
228 seconds were coded as such. It leaded to damn rounding errors.
229 - Bugfix: Hard to belive that timers ever worked before this.
231 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
233 SimGrid (2.91) unstable; urgency=low
235 Alpha 2 on the path to SimGrid 3
239 - Bug fix in the lmm_solver.
241 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
242 through the function MSG_paje_output.
243 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
244 - It is possible to bound the rate of a communication in MSG with
245 MSG_task_put_bounded() (was already in the previous version but I had forgotten
246 to write it in the changelog).
247 - Bug fix to let GRAS run on top of MSG until we move it directly on top
252 - Various cleanups to the autotools stuff
253 - Begin to move Gras examples to examples/gras/
254 - Let make distcheck work again (yeah!)
256 - documentation overhauled using doxygen.
257 gtk-doc-tools is dead in SimGrid now.
258 - Automatically extract all existing logging categories, and add the list
259 to the documentation (long standing one, to say the less)
261 - Cleanup the known architecture table. Reorder the entries to group what
262 should be, and use a more consistent naming scheme.
263 (some of the test dataset are still to be regenerated)
264 - New! Allow library to register globals on each process just as userdata
266 This is implemented using a xbt_dict and not a xbt_set, so we loose the
267 lookup time (for now).
268 Use it in msg and trp.
269 This cleans a lot the internals and helps enforcing privacy of the
270 headers between the gras components.
271 - New! Add a timer mechanism, not unlike cron(8) and at(1).
272 - Bugfix: gras_os_time was delirious in RL.
273 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
275 - Reenable GRAS now that it works.
277 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
279 SimGrid (2.90) unstable; urgency=low
281 Alpha 1 on the path to SimGrid 3
283 * It is a long time since the last release of SimGrid. I'm sorry about
284 that but as I had told you, I was rewriting a lot of things. I apologize
285 to those who had been reporting bugs to me and that I had not answered.
286 If your bug is still in the new version, please tell me. Here is a
287 summary of the main changes.
289 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
290 lead by Martin Quinson. As a consequence SimGrid gains a lot in
291 portability, speed, and a lot more but you'll figure it out later.
292 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
293 I wanted to release the new MSG as soon as possible and I have
294 broken GRAS, which is the reason why, for now, only MSG is fully
295 functional. A laconic description of these projects is available
296 in the documentation.
298 * REVOLUTION 2: I have removed SG and I am now using a new simulation
299 kernel optimized for our needs (called SURF but only the developers
300 should use it). Hence, MSG is now roughly 30 times faster and I think
301 that by rewriting a little bit MSG, I could event speed it up a little
302 bit more. Beside the gain in speed, it is also much easier to encode a
303 new platform model with SURF than it was with SG. More to come...
305 * REVOLUTION 3: I have tried to change a little as possible the API of
306 MSG but a few things really had to disappear. The main differences
307 with the previous version are :
308 1) no more m_links_t and the corresponding functions. Platforms are
309 directly read from a XML description and cannot be hard-coded
310 anymore. The same format is used for application deployment
311 description. The new format is described in the documentation.
312 Have a look in tools/platform_generation. There is a tiny script
313 that converts from the old platform format to the new one. Concerning
314 the application deployment format, parsing the old one is tricky.
315 I think most of you should however be able to convert your files. If
316 it is really an issue, I can write a C code that does the conversion.
318 2) the toolbox tbx does not exist anymore. We now have a library
319 with much more data-structures but without the hash-tables (we have
320 dictionaries that are much faster).
322 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
324 *****************************************************************************
325 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
326 * don't feel like converting the oldies. (Mt) *
327 *****************************************************************************
330 Version 2.90: "the long awaited one"
331 - Finished rewriting and debugging MSG. Rewrote the documentation.
332 - disable GRAS for now since it needs to be ported to the newest SG
335 - Finish the port to windows (using mingw32 for cross-compile)
338 - Main loop and datastructures of SURF. A cpu resource object is
339 functional. Surf can thus be used to create cpu's with variable
340 performance on which you can execute some actions.
342 2004-11-15 Martin Quinson
343 - Port to ARM. Simply added the alignment and size descriptions. Should
344 work, but the ARM machines are so slow that I didn't had the opportunity
345 to 'make check' over there yet.
347 2004-11-15 Arnaud Legrand
348 - Trace manager now written. It uses a heap structure and is therefore
349 expected to be efficient. It may however be speeded up (particularly
350 when many events occur at the same date) by using red and black
351 trees. One day maybe...
352 - Max-min linear system solver written. It uses a sparse matrix
353 structure taking advantage of its expected use. Most operations are
354 O(1) and free/calloc are called as few as possible. The computation of
355 the minimum could however be improved by using a red and black tree
358 2004-11-03 Arnaud Legrand
359 - Rename every gras_* function that was in xbt/ to its xbt_
361 - Add a heap and a doubly-linked list to xbt
362 - Added a dichotomy to the dictionaries. make check works as well before
363 so I assume that the patch is correct. I do not know however if things
364 run effectively faster than before now. :)
366 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
367 SimGrid, and the version number is bumped to 2.x
369 2004-10-29 Martin Quinson
370 - Introduction of the remote errors.
371 They are the result of a RMI/RPC on the remote machine.
372 ErrCodes being scalar values, you can't get the host on which those
373 errors did happen. Extending the error mechanism as in Gnome is possible.
374 No idea yet whether it is a good idea.
376 2004-10-28 Martin Quinson
377 - Interface revolution: the Starred Structure Eradication.
378 I used to do typedef struct {} toto_t; and then handle *toto_t.
379 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
380 - ???_t is a valid type (builded with typedef)
381 - s_toto_t is a structure (access to fields with .)
382 - s_toto is a structure needing 'struct' keyword to be used
383 - e_toto_t is an enum
384 - toto_t is an 'object' (struct*)
386 typedef struct s_toto {} s_toto_t, *toto_t;
387 typedef enum {} e_toto_t;
388 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
389 s_toto_t) is private.
391 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
392 so that it can set the variable to NULL. It was so for dicts and sets,
393 it changed for dynars.
395 - Fix a bunch of memleaks in dict_remove
396 - Fix a bug in sg/server_socket opening: it failed all the time.
398 2004-10-07 Martin Quinson
399 - Speed up dynar lookup operation a bit.
401 gras_dynar_get is dead.
403 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
404 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
405 (which gives you the address of the stored data).
407 gras_dynar_get_as is an helpful macro which allows you to retrieve a
408 copy of the data using an affectation to do the job and not a memcpy.
410 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
411 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
413 It does not really speedup the dynar test because they are
414 setting elements all the time (and look them seldom). But the dict does
415 far more lookup than setting.
417 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
419 2004-10-05 Martin Quinson
420 - Allow to (en/dis)able the cycle detection at run time.
422 Whether we should check for cycle or not is now a property of each
423 datatype. When you think there may be some cycle, use datadesc_cycle_set.
424 datadesc_cycle_unset allow to remove this property when previously set.
426 Note that the cycle detection is off by default since it impacts the
427 performance. Watch the data you feed GRAS with ;)
429 This property is hereditary. Any element embedded in a structure having it
430 set have it set for the time of this data exchange.
432 You should set it both on sender and receiver side. If you don't set it on
433 sender side, it will enter an endless loop. If you forget on receiver
434 side, the cycles won't be recreated after communication.
436 - Header reorganization.
437 Kill gras_private.h, each submodule must load the headers it needs.
439 2004-10-04 Martin Quinson
440 - Interface revolution: do not try to survive to malloc failure.
442 Now, gras_malloc and friends call gras_abort() on failure.
443 As a conclusion, malloc_error is not a valid error anymore, and all
444 functions for which it was the only gras_error_t return value are
445 changed. They now return void, or there result directly.
446 This simplify the API a lot.
448 2004-09-29 Martin Quinson
449 - Re-enable raw sockets.
450 Created by gras_socket_{client,server}_ext;
451 Used with gras_raw_{send,recv}
454 It should allow to kill the last bits of gras first version soon.
456 This is not completely satisfactory yet (duplicate code with
457 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
460 - Simplify transport plugin (internal) interface by not passing any
461 argument to _server and _client, but embedding them in the socket
464 2004-09-28 Martin Quinson
465 - Finish the port to AIX.
466 autoconf was my problem (segfault within the malloc replacement
467 function. No idea why)
469 2004-09-16 Martin Quinson
470 - Fix some size_t madness on 64bit architectures.
472 2004-09-08 Martin Quinson
473 - Reduce the number of system headers loaded, overload some more system
474 calls (such as malloc to cast the result of the system one, and work
476 - Fix and reintroduce the config support
478 2004-09-07 Martin Quinson
479 - Source code reorganization to allow Arnaud to surf all over there.
480 - Allow to document the logging categories.
481 - Remove all uppercase from logging categories and useless cleanup in names.
483 2004-08-18 Martin Quinson
484 Version 0.6.2 (protocol not changed; API changed)
485 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
486 gras_error_t), and NULL when not found. Functions expecting a msgtype
487 as argument (msg_wait; msg_send) deal with NULL argument by providing a
488 hopefully usefull message.
489 - Portability to prehistoric sparcs again
491 2004-08-17 Martin Quinson
492 Version 0.6.1 (protocol not changed; ABI not changed)
493 - prealloc some buffers to speed things up
495 2004-08-11 Martin Quinson
496 Version 0.6 (protocol not changed; ABI expended)
497 - The parsing macro can deal with the references, provided that you add
498 the relevant annotations (using GRAS_ANNOTE(size,field_name))
500 2004-08-09 Martin Quinson
501 Version 0.5 (protocol not changed; ABI changed)
502 - Allow to off turn the cycle detection code in data exchange at
503 compilation time. It should be at run time, but I'm short of time (and
504 the config stuff is still broken). That way, we keep dict out of the
505 critical path, which is good because the performance is poor:
506 - search not dichotomial yet
507 - dynar give no way to access their content and memcpy everytime
508 - In composed data description (struct, ref and so on), stop foolness of
509 keeping the subtype's ID, but store the type itself. This keeps sets out
510 of the critical path, which is good since they rely on dynar and
511 dictionnaries. The only loose of that is that we cannot detect the
512 redeclaration of a structure/union with another content (but I'm not sure
513 the code detected well this error before anyway). We still can detect
514 the redefinition discrepancy for the other types.
515 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
516 it breaks the code because of type-punning used all over the place).
517 This breaks on all non-gcc architectures (for now).
519 All those changes (plus the buffer of last time) allow me to gain 2 order
520 of magnitude on cruel tests consisting of 800000 array of integers on two
521 level of a hierarchical structure (200 secondes -> 4 secondes)
524 - the selector of reference must now return the type it points to, not
527 2004-08-06 Martin Quinson
528 Version 0.4 (protocol changed; ABI not changed)
529 - Allow to pass --gras-log argument to processes in simulation mode. Really.
530 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
531 - Implement a buffer transport, and use it by default (it relies on tcp in
532 real life and on sg in simulation).
533 That's a bit hackish since I had a new field to the structure to store
534 its data without interfering with the subtype ones. Inheritance
535 is tricky in C. And that's a kind of reverse inheritance with one class
536 derivating two classes. Or maybe a game with java interfaces. Anyway,
537 that's damn hard in C (at least).
538 Moreover, I got tired while trying to ensure plugin separation and
539 genericity in SG mode. MSG wants me to do weird things, so let's go for
540 cruel hacks (temporarily of course ;).
541 See comment in transport_private.h:71
542 - do not include all the _interface headers in private but in the files
543 which really need them (to cut the compilation time when they are
546 2004-07-26 Martin Quinson
547 Version 0.3 (protocol not changed; ABI changed)
548 - Major overhault of the datadesc interface to simplify it:
549 - shorted the function names:
550 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
551 - add a trivial way to push/pop integers into the cbps without malloc.
552 This allows to make really generic sub_type description, which simply
553 pop their size of the stack.
554 - add a function gras_datadesc_ref_pop_arr() which does what users want
555 most of the time: Declare a dynamic array (which pops its size of the
556 stack) and declare a reference to it. Poor name, but anyway.
557 - kill the post-send callback, add a post-receive one
559 2004-07-23 Martin Quinson
560 Version 0.2 (protocol changed; ABI changed)
561 - add some testing for cpbs in the test cases, and fix some more bugs.
562 This invalidate again the little64 data file, since I cannot regenerate
564 - remove an awfull optimization in the logging stuff, allowing me to:
565 - understand it again
566 - learn gcc how to check that the argument match the provided format
567 - fix all errors revealed by gcc after that
568 - internal keys of dict are not \0 terminated. Deal with it properly in
569 loggings instead of segfaulting when the user want to see the logs :-/
571 2004-07-22 Martin Quinson
572 - Fix some stupid bug preventing cbps (callback postit) from working
574 2004-07-21 Martin Quinson
575 - Some documentation cleanups
576 - remove the useless last argument of msgtype_declare
577 - rename the Virtu functions to fit into the 'os' namespace
578 - move headers src/include -> src/include/gras/ and stop fooling with
579 gras -> . symbolic link
580 - make distcheck is now successful
582 2004-07-19 Martin Quinson
584 - Build shared library also
585 - Install html doc to the right location
586 - stop removing maintainer files in make clean
587 - build tests only on make check
589 2004-07-13 Martin Quinson
591 - No major issue in previous version => change versionning schema
592 - Re-enable little64 convertion test now that Abdou kindly regenerated the
593 corresponding dataset.
595 2004-07-11 Martin Quinson
597 - Get it working with any kind of structure (we can compute the padding
598 bytes remotely for all the architectures I have access to)
599 - Implement the structure parsing macro (still not quite robust/complete)
600 - Improvement to the remote testing toysuite
602 2004-07-10 Martin Quinson
604 - get ride of a bunch of deprecated macros
605 - actually run the test for two-compliment, not only compile it :-/
606 - test whether the structures get packed (and bail out if yes. Damn.
607 Alignment is a serious matter)
608 - test whether the structures get compacted (but respecting the alignment
609 constraints of each types)
610 - test whether the array fields of structures can straddle alignment boundaries
612 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
613 (took me 2 hours to find that bug, looking at the wrong place)
614 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
615 before sending/receiving and are used to compute the offsets of fields
616 - Given that padding size depend even on compiler options, keep track of
617 alignment and aligned_size only for the current architecture. Not a big
618 deal since we send structure fields one after the other (seems
620 - Add the datastructure used for IEEE paper by the PBIO guys to the test
621 program, let it work on linux/gcc/little32. portability todo.
623 2004-07-08 Martin Quinson
624 - import and improve remote compilation support from FAST
625 - make sure make check works on half a dozen of machines out there
627 2004-07-07 Martin Quinson
628 Let's say it's version 0.0.3 ;)
629 - Implement conversions (yuhu!)
630 - Let it work on solaris (beside conversion, of course)
631 - Stupid me, using rand() to generate the conversion datatests in not wise.
633 2004-07-06 Martin Quinson
634 - Let make dist work, since I'm gonna need it to compile on remote hosts
635 - Let Tests/datadesc_usage write the architecture on which the file was
636 generated as first byte.
637 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
638 architecture descriptions.
639 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
640 on those architectures.
641 - Optimization: send/recv array of scalar in one shoot
643 2004-07-05 Martin Quinson
644 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
646 - Plug a whole bunch of memleaks
647 - each process now have to call gras_{init,exit}. One day, their log
648 settings will be separated
649 - Continue the code factorisation between SG, RL and common in Transport.
651 2004-07-04 Martin Quinson
653 - Redistribution between SG and RL.
654 We wanna have to accept in SG, so move accepted related parts of RL in
655 the common part. (more precisely, the dynar of all known sockets is no
656 more a static in transport.c, but part of the process_data)
658 [gras_stub_generator]
659 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
661 2004-07-03 Martin Quinson
662 - Create a new log channel tbx containing dict, set, log, dynar (to shut
663 them all up in one shot)
665 - Fix the ugly case of reference to dynamic array.
666 - New (semi-public) function gras_datadesc_size to allow the messaging
667 layer to malloc the needed space for the buffer.
669 - gras_socket_close now expect the socket to close (and not its address to
670 put NULL in it after it). This is because the socket passed to handlers
671 is one of their argument (=> not writable).
673 - propagate the interface cleanup from last week in datadesc, ie remove a
674 superfluous level of indirection. User pass adress of variable
675 containing data (both when sending and receiving), and not of a variable
676 being a pointer to the data. Let's say that I like it better ;)
677 The price for that is constructs like "int msg=*(int*)payload" in
678 handlers, but it's a fine price, IMHO.
680 - Let it work in RL (yuhu)
682 2004-06-21 Martin Quinson
684 - porting SG plugin and SG select to new standards (works almost).
685 - plug memleaks and fix bugs around.
688 - cleanup the prototype of data recv and force users to specify when they
689 want to handle references to objects. Test case working even for cycles.
690 - plug memleaks. Valgrind is perfectly ok with this.
692 2004-06-12 Martin Quinson
694 - cleanup the separation between plugin and main code in plugin creation
696 2004-06-11 Martin Quinson
698 - Reput hook for raw sockets, needed for BW experiments
699 - kill a few lines of dead code
700 [Data description] Interface cleanup
701 - gras_datadesc_by_name returns the searched type or NULL.
702 That way, no variable is needed to use a type desc once, which makes
704 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
705 two parameters were strings (field name, type name), leading to
707 [Dicos] Interface cleanup
708 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
709 This is consistant with the dynar API.
711 2004-04-21 Martin Quinson
713 - Porting to new standards.
716 There is no bag anymore, no need to take extra provision to mask the
717 pointers behind "ID".
718 Better splitup of functions between files create/exchange/convert.
719 This is still a bit artificial since convert and receive are so
720 interleaved, but anyway.
722 - add a queued message list to procdata (the ones not matching criteria
724 - factorize some more code between SG and RL wrt procdata
726 - use gras_exit in example to track memleaks
727 - get rid of gs_example now that GS is properly integrated into gras
728 - update run_test to integrate the lastest tests (datadesc)
730 - rename WARNINGn macros to WARNn since it prooved error-prone
732 2004-04-19 Martin Quinson
734 - register init/exit functions within gras module mechanism
735 - send/receive function.
736 Convertion is not implemented, but short-cutted if not needed.
737 struct/array elements are sent one by one (instead of block-wise), but
738 nobody really cares (yet). Get a prototype before optimizing.
739 - tests (using a file socket) for DD send/receive on:
740 - base types: int, float
741 - array: fixed size, string (ie ref to dynamic string)
742 - structure: homogeneous, heterogeneous
743 - chained list, graph with cycle
744 Believe it or not, valgrind is not too unhappy with the results. The
745 cycle happily segfaults, but the others are ok. And I'm sick of pointers
749 - Bugfix when using a filename explicitely (instead of '-')
751 2004-04-09 Martin Quinson
753 - factorize more code between RL and SG in socket creation
754 - Complete the implementation and tests of:
756 o file (only in RL, and mainly for debugging)
758 I lost 3 days to design a portable address resolver, and then decided
759 that the prototype mainly have to run on my box.
760 Addressing portability too early may be like optimizing too early :-/
762 - use gras_init in the Tests instead of the crappy parse_log_opt
763 (the latter function is removed)
764 [Conditional execution]
765 - New functions: gras_if_RL/gras_if_SG (basic support for this)
766 [Code reorganisation]
767 - Get rid of libgrasutils.a since it makes more trouble than it solves.
768 Build examples against the RL library, since there is no way to disable
769 its creation for now.