1 SimGrid (2.96) unstable; urgency=low
4 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
5 This deprecates the xbt_error_t mecanisms.
6 It modifies (simplifies) all XBT and GRAS API.
7 MSG API keeps unchanged (exceptions raised by XBT are catched from
8 within MSG and masked with existing error handling facilities)
11 * New! Add a FATPIPE model. [AL]
12 * New! Add a parallel task model. [AL]
13 * New! Add automatically a loopback interface (in the default
14 network model) if none was precised.
17 * Bugfix: MSG_process_resume now works with the current running process.
19 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
20 * Modification of MSG_task_get_compute_duration. Once a task has been
21 processed, the value returned by this function is now equal to 0. [AL]
22 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
23 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
24 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
25 of a m_task. For now, MSG_task_cancel only works with computation
27 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
28 of the processor (in Mflop/s) regardless of the current load on the
33 SimGrid (2.95) unstable; urgency=low
36 * Steal some nice code to GNU pth to fix context detection and usage [AL]
37 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
38 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
41 * Bugfix: Allow absolute paths for platform description files [MQ]
42 * Bugfix: do free the variables after use. Leads to drastic performance
44 * Implement max_duration (ie, timeouts) on resources [AL]
47 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
49 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
50 the needed support (timeouts) [AL]
53 * Implement measurement sockets. You can now get the bandwidth between two
54 hosts thanks to AMOK (see below). [MQ]
55 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
56 dynar over the network (yeah) [MQ]
57 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
58 * simple/static token-ring example. [Alexandre Colucci and MQ]
59 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
63 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
64 type parsing. "short" and "long" modifiers were also ignored; other
65 modifier (such as reference level) are still ignored. [MQ]
66 * Embeed the buffer size within the buffer itself on SG. [MQ]
67 That way, send() are atomic and cannot get intermixed anymore (at least
68 the ones which are less than 100k; bigger messages still have the issue)
69 * Array size pushed by the field, not by the field type (or each
70 and every long int will push stuff to the cbps) [MQ]
71 * use select() to sleep since it allows to portably sleep less than one
75 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
78 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
79 * type category 'ignored' killed as it was never used and were difficult
81 * whether a type can cycle or not is now a flag, leaving room for more
82 flags (as "ignored", if we feel the need one day ;)
83 * Rename raw sockets to measurement sockets since "raw" has another
84 meaning in networking community.
87 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
88 and offers features not belonging to GRAS but that most applications
89 need. One day, it may be a set of plugins loadable at runtime.
90 * New module: bandwidth
91 bandwidth measurement between arbitrary nodes running this module. [MQ]
93 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
95 SimGrid (2.94) unstable; urgency=low
97 The first beta release of SimGrid 3 !
101 * Update the main page and the FAQ. Adding references to gforge.
104 * Add a gras_os_getpid function.
107 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
108 * Extend the logs so that they also print PID, hostname, date, ... if
110 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
111 and kill the old version which were in testsuite/
112 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
116 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
117 for users. Instead, each programming environment (GRAS, MSG,...) use it
118 when needed to provide such a feature to users.
119 Don't shortcut the mecanism or you will also shortcut the virtualization
120 you need on the simulator.
124 * Cleanups in configury with regard to compile optimization/warning flags.
125 Also add -fno-loop-optimize to any powerpc since it's the optimization
126 killing gcc (< 3.4.0).
127 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
128 needed by MSG examples complications
129 * Borrow configury beautifications from PHP
132 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
133 warning (thanks loris for stumbling into this one).
134 * Bugfix: stop loading private headers (gras_config.h) from the public
138 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
139 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
140 RL would imply a DNS resolver, which is *hard* to do in a portable way
141 (and therefore delayed).
142 * Implement a real timer mecanism and use it in timing macros. This allows
143 to avoid rounding errors and get a 0.000005 sec precision in timing
144 macros. While I was at it, various cleanups:
145 - allow to declare more than one timed section per file (fix a stupid bug)
146 - move some private declaration to the right place
147 - merge conditional execution and timing macros into emulation module
148 - document the module
149 - make sure the module cleanups its mess on gras_exit
150 * Documentation improvements:
151 - (new) how to compile applications using GRAS
152 - (new) emulation support (timing macros)
154 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
156 SimGrid (2.93) unstable; urgency=low
158 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
161 - Use Paje properly where used. Still to be sanitized properly.
162 - Portability fix: Add an implementation of the contexts using pthread
166 - Add xbt_procname(): returns the name of the current process.
167 Use it to show the current process's name in all logging.
169 - fix detection of older flex version and the reaction, since we do
170 depend on modern ones (we use lex_destroy)
171 - Better separation of SG and RL in the libs: remove all simulation code
172 from libgras. As a result, this lib is now only 200k when stripped.
173 Some of the xbt modules may also be duplicated (two sets and such) and
174 should be cleaned/killed before SG3.
175 - Insist on using xlC on AIX because of weird problems involving gcc there.
176 - Cleanup the make remote stuff. This is now done by scripts
177 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
178 mainly for our private use, but we're working on changing them to user
181 - Bugfix: flush the socket on close only if there is some *output*.
182 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
184 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
187 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
189 SimGrid (2.92) unstable; urgency=low
191 Alpha 3 on the path to SimGrid 3
195 - New! First try of benchmarking macros.
196 - New! First try so that gras_stub_generator generate deployment and
197 remote compilation helpers.
199 - Bugfix: Initialization fix in msg_test.
203 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
205 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
207 - API Change: gras_os_sleep to take the amount of seconds as a double.
208 Accepting an int was error prone since it was the only location where
209 seconds were coded as such. It leaded to damn rounding errors.
210 - Bugfix: Hard to belive that timers ever worked before this.
212 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
214 SimGrid (2.91) unstable; urgency=low
216 Alpha 2 on the path to SimGrid 3
220 - Bug fix in the lmm_solver.
222 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
223 through the function MSG_paje_output.
224 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
225 - It is possible to bound the rate of a communication in MSG with
226 MSG_task_put_bounded() (was already in the previous version but I had forgotten
227 to write it in the changelog).
228 - Bug fix to let GRAS run on top of MSG until we move it directly on top
233 - Various cleanups to the autotools stuff
234 - Begin to move Gras examples to examples/gras/
235 - Let make distcheck work again (yeah!)
237 - documentation overhauled using doxygen.
238 gtk-doc-tools is dead in SimGrid now.
239 - Automatically extract all existing logging categories, and add the list
240 to the documentation (long standing one, to say the less)
242 - Cleanup the known architecture table. Reorder the entries to group what
243 should be, and use a more consistent naming scheme.
244 (some of the test dataset are still to be regenerated)
245 - New! Allow library to register globals on each process just as userdata
247 This is implemented using a xbt_dict and not a xbt_set, so we loose the
248 lookup time (for now).
249 Use it in msg and trp.
250 This cleans a lot the internals and helps enforcing privacy of the
251 headers between the gras components.
252 - New! Add a timer mechanism, not unlike cron(8) and at(1).
253 - Bugfix: gras_os_time was delirious in RL.
254 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
256 - Reenable GRAS now that it works.
258 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
260 SimGrid (2.90) unstable; urgency=low
262 Alpha 1 on the path to SimGrid 3
264 * It is a long time since the last release of SimGrid. I'm sorry about
265 that but as I had told you, I was rewriting a lot of things. I apologize
266 to those who had been reporting bugs to me and that I had not answered.
267 If your bug is still in the new version, please tell me. Here is a
268 summary of the main changes.
270 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
271 lead by Martin Quinson. As a consequence SimGrid gains a lot in
272 portability, speed, and a lot more but you'll figure it out later.
273 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
274 I wanted to release the new MSG as soon as possible and I have
275 broken GRAS, which is the reason why, for now, only MSG is fully
276 functional. A laconic description of these projects is available
277 in the documentation.
279 * REVOLUTION 2: I have removed SG and I am now using a new simulation
280 kernel optimized for our needs (called SURF but only the developers
281 should use it). Hence, MSG is now roughly 30 times faster and I think
282 that by rewriting a little bit MSG, I could event speed it up a little
283 bit more. Beside the gain in speed, it is also much easier to encode a
284 new platform model with SURF than it was with SG. More to come...
286 * REVOLUTION 3: I have tried to change a little as possible the API of
287 MSG but a few things really had to disappear. The main differences
288 with the previous version are :
289 1) no more m_links_t and the corresponding functions. Platforms are
290 directly read from a XML description and cannot be hard-coded
291 anymore. The same format is used for application deployment
292 description. The new format is described in the documentation.
293 Have a look in tools/platform_generation. There is a tiny script
294 that converts from the old platform format to the new one. Concerning
295 the application deployment format, parsing the old one is tricky.
296 I think most of you should however be able to convert your files. If
297 it is really an issue, I can write a C code that does the conversion.
299 2) the toolbox tbx does not exist anymore. We now have a library
300 with much more data-structures but without the hash-tables (we have
301 dictionaries that are much faster).
303 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
305 *****************************************************************************
306 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
307 * don't feel like converting the oldies. (Mt) *
308 *****************************************************************************
311 Version 2.90: "the long awaited one"
312 - Finished rewriting and debugging MSG. Rewrote the documentation.
313 - disable GRAS for now since it needs to be ported to the newest SG
316 - Finish the port to windows (using mingw32 for cross-compile)
319 - Main loop and datastructures of SURF. A cpu resource object is
320 functional. Surf can thus be used to create cpu's with variable
321 performance on which you can execute some actions.
323 2004-11-15 Martin Quinson
324 - Port to ARM. Simply added the alignment and size descriptions. Should
325 work, but the ARM machines are so slow that I didn't had the opportunity
326 to 'make check' over there yet.
328 2004-11-15 Arnaud Legrand
329 - Trace manager now written. It uses a heap structure and is therefore
330 expected to be efficient. It may however be speeded up (particularly
331 when many events occur at the same date) by using red and black
332 trees. One day maybe...
333 - Max-min linear system solver written. It uses a sparse matrix
334 structure taking advantage of its expected use. Most operations are
335 O(1) and free/calloc are called as few as possible. The computation of
336 the minimum could however be improved by using a red and black tree
339 2004-11-03 Arnaud Legrand
340 - Rename every gras_* function that was in xbt/ to its xbt_
342 - Add a heap and a doubly-linked list to xbt
343 - Added a dichotomy to the dictionaries. make check works as well before
344 so I assume that the patch is correct. I do not know however if things
345 run effectively faster than before now. :)
347 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
348 SimGrid, and the version number is bumped to 2.x
350 2004-10-29 Martin Quinson
351 - Introduction of the remote errors.
352 They are the result of a RMI/RPC on the remote machine.
353 ErrCodes being scalar values, you can't get the host on which those
354 errors did happen. Extending the error mechanism as in Gnome is possible.
355 No idea yet whether it is a good idea.
357 2004-10-28 Martin Quinson
358 - Interface revolution: the Starred Structure Eradication.
359 I used to do typedef struct {} toto_t; and then handle *toto_t.
360 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
361 - ???_t is a valid type (builded with typedef)
362 - s_toto_t is a structure (access to fields with .)
363 - s_toto is a structure needing 'struct' keyword to be used
364 - e_toto_t is an enum
365 - toto_t is an 'object' (struct*)
367 typedef struct s_toto {} s_toto_t, *toto_t;
368 typedef enum {} e_toto_t;
369 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
370 s_toto_t) is private.
372 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
373 so that it can set the variable to NULL. It was so for dicts and sets,
374 it changed for dynars.
376 - Fix a bunch of memleaks in dict_remove
377 - Fix a bug in sg/server_socket opening: it failed all the time.
379 2004-10-07 Martin Quinson
380 - Speed up dynar lookup operation a bit.
382 gras_dynar_get is dead.
384 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
385 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
386 (which gives you the address of the stored data).
388 gras_dynar_get_as is an helpful macro which allows you to retrieve a
389 copy of the data using an affectation to do the job and not a memcpy.
391 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
392 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
394 It does not really speedup the dynar test because they are
395 setting elements all the time (and look them seldom). But the dict does
396 far more lookup than setting.
398 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
400 2004-10-05 Martin Quinson
401 - Allow to (en/dis)able the cycle detection at run time.
403 Whether we should check for cycle or not is now a property of each
404 datatype. When you think there may be some cycle, use datadesc_cycle_set.
405 datadesc_cycle_unset allow to remove this property when previously set.
407 Note that the cycle detection is off by default since it impacts the
408 performance. Watch the data you feed GRAS with ;)
410 This property is hereditary. Any element embedded in a structure having it
411 set have it set for the time of this data exchange.
413 You should set it both on sender and receiver side. If you don't set it on
414 sender side, it will enter an endless loop. If you forget on receiver
415 side, the cycles won't be recreated after communication.
417 - Header reorganization.
418 Kill gras_private.h, each submodule must load the headers it needs.
420 2004-10-04 Martin Quinson
421 - Interface revolution: do not try to survive to malloc failure.
423 Now, gras_malloc and friends call gras_abort() on failure.
424 As a conclusion, malloc_error is not a valid error anymore, and all
425 functions for which it was the only gras_error_t return value are
426 changed. They now return void, or there result directly.
427 This simplify the API a lot.
429 2004-09-29 Martin Quinson
430 - Re-enable raw sockets.
431 Created by gras_socket_{client,server}_ext;
432 Used with gras_raw_{send,recv}
435 It should allow to kill the last bits of gras first version soon.
437 This is not completely satisfactory yet (duplicate code with
438 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
441 - Simplify transport plugin (internal) interface by not passing any
442 argument to _server and _client, but embedding them in the socket
445 2004-09-28 Martin Quinson
446 - Finish the port to AIX.
447 autoconf was my problem (segfault within the malloc replacement
448 function. No idea why)
450 2004-09-16 Martin Quinson
451 - Fix some size_t madness on 64bit architectures.
453 2004-09-08 Martin Quinson
454 - Reduce the number of system headers loaded, overload some more system
455 calls (such as malloc to cast the result of the system one, and work
457 - Fix and reintroduce the config support
459 2004-09-07 Martin Quinson
460 - Source code reorganization to allow Arnaud to surf all over there.
461 - Allow to document the logging categories.
462 - Remove all uppercase from logging categories and useless cleanup in names.
464 2004-08-18 Martin Quinson
465 Version 0.6.2 (protocol not changed; API changed)
466 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
467 gras_error_t), and NULL when not found. Functions expecting a msgtype
468 as argument (msg_wait; msg_send) deal with NULL argument by providing a
469 hopefully usefull message.
470 - Portability to prehistoric sparcs again
472 2004-08-17 Martin Quinson
473 Version 0.6.1 (protocol not changed; ABI not changed)
474 - prealloc some buffers to speed things up
476 2004-08-11 Martin Quinson
477 Version 0.6 (protocol not changed; ABI expended)
478 - The parsing macro can deal with the references, provided that you add
479 the relevant annotations (using GRAS_ANNOTE(size,field_name))
481 2004-08-09 Martin Quinson
482 Version 0.5 (protocol not changed; ABI changed)
483 - Allow to off turn the cycle detection code in data exchange at
484 compilation time. It should be at run time, but I'm short of time (and
485 the config stuff is still broken). That way, we keep dict out of the
486 critical path, which is good because the performance is poor:
487 - search not dichotomial yet
488 - dynar give no way to access their content and memcpy everytime
489 - In composed data description (struct, ref and so on), stop foolness of
490 keeping the subtype's ID, but store the type itself. This keeps sets out
491 of the critical path, which is good since they rely on dynar and
492 dictionnaries. The only loose of that is that we cannot detect the
493 redeclaration of a structure/union with another content (but I'm not sure
494 the code detected well this error before anyway). We still can detect
495 the redefinition discrepancy for the other types.
496 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
497 it breaks the code because of type-punning used all over the place).
498 This breaks on all non-gcc architectures (for now).
500 All those changes (plus the buffer of last time) allow me to gain 2 order
501 of magnitude on cruel tests consisting of 800000 array of integers on two
502 level of a hierarchical structure (200 secondes -> 4 secondes)
505 - the selector of reference must now return the type it points to, not
508 2004-08-06 Martin Quinson
509 Version 0.4 (protocol changed; ABI not changed)
510 - Allow to pass --gras-log argument to processes in simulation mode. Really.
511 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
512 - Implement a buffer transport, and use it by default (it relies on tcp in
513 real life and on sg in simulation).
514 That's a bit hackish since I had a new field to the structure to store
515 its data without interfering with the subtype ones. Inheritance
516 is tricky in C. And that's a kind of reverse inheritance with one class
517 derivating two classes. Or maybe a game with java interfaces. Anyway,
518 that's damn hard in C (at least).
519 Moreover, I got tired while trying to ensure plugin separation and
520 genericity in SG mode. MSG wants me to do weird things, so let's go for
521 cruel hacks (temporarily of course ;).
522 See comment in transport_private.h:71
523 - do not include all the _interface headers in private but in the files
524 which really need them (to cut the compilation time when they are
527 2004-07-26 Martin Quinson
528 Version 0.3 (protocol not changed; ABI changed)
529 - Major overhault of the datadesc interface to simplify it:
530 - shorted the function names:
531 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
532 - add a trivial way to push/pop integers into the cbps without malloc.
533 This allows to make really generic sub_type description, which simply
534 pop their size of the stack.
535 - add a function gras_datadesc_ref_pop_arr() which does what users want
536 most of the time: Declare a dynamic array (which pops its size of the
537 stack) and declare a reference to it. Poor name, but anyway.
538 - kill the post-send callback, add a post-receive one
540 2004-07-23 Martin Quinson
541 Version 0.2 (protocol changed; ABI changed)
542 - add some testing for cpbs in the test cases, and fix some more bugs.
543 This invalidate again the little64 data file, since I cannot regenerate
545 - remove an awfull optimization in the logging stuff, allowing me to:
546 - understand it again
547 - learn gcc how to check that the argument match the provided format
548 - fix all errors revealed by gcc after that
549 - internal keys of dict are not \0 terminated. Deal with it properly in
550 loggings instead of segfaulting when the user want to see the logs :-/
552 2004-07-22 Martin Quinson
553 - Fix some stupid bug preventing cbps (callback postit) from working
555 2004-07-21 Martin Quinson
556 - Some documentation cleanups
557 - remove the useless last argument of msgtype_declare
558 - rename the Virtu functions to fit into the 'os' namespace
559 - move headers src/include -> src/include/gras/ and stop fooling with
560 gras -> . symbolic link
561 - make distcheck is now successful
563 2004-07-19 Martin Quinson
565 - Build shared library also
566 - Install html doc to the right location
567 - stop removing maintainer files in make clean
568 - build tests only on make check
570 2004-07-13 Martin Quinson
572 - No major issue in previous version => change versionning schema
573 - Re-enable little64 convertion test now that Abdou kindly regenerated the
574 corresponding dataset.
576 2004-07-11 Martin Quinson
578 - Get it working with any kind of structure (we can compute the padding
579 bytes remotely for all the architectures I have access to)
580 - Implement the structure parsing macro (still not quite robust/complete)
581 - Improvement to the remote testing toysuite
583 2004-07-10 Martin Quinson
585 - get ride of a bunch of deprecated macros
586 - actually run the test for two-compliment, not only compile it :-/
587 - test whether the structures get packed (and bail out if yes. Damn.
588 Alignment is a serious matter)
589 - test whether the structures get compacted (but respecting the alignment
590 constraints of each types)
591 - test whether the array fields of structures can straddle alignment boundaries
593 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
594 (took me 2 hours to find that bug, looking at the wrong place)
595 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
596 before sending/receiving and are used to compute the offsets of fields
597 - Given that padding size depend even on compiler options, keep track of
598 alignment and aligned_size only for the current architecture. Not a big
599 deal since we send structure fields one after the other (seems
601 - Add the datastructure used for IEEE paper by the PBIO guys to the test
602 program, let it work on linux/gcc/little32. portability todo.
604 2004-07-08 Martin Quinson
605 - import and improve remote compilation support from FAST
606 - make sure make check works on half a dozen of machines out there
608 2004-07-07 Martin Quinson
609 Let's say it's version 0.0.3 ;)
610 - Implement conversions (yuhu!)
611 - Let it work on solaris (beside conversion, of course)
612 - Stupid me, using rand() to generate the conversion datatests in not wise.
614 2004-07-06 Martin Quinson
615 - Let make dist work, since I'm gonna need it to compile on remote hosts
616 - Let Tests/datadesc_usage write the architecture on which the file was
617 generated as first byte.
618 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
619 architecture descriptions.
620 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
621 on those architectures.
622 - Optimization: send/recv array of scalar in one shoot
624 2004-07-05 Martin Quinson
625 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
627 - Plug a whole bunch of memleaks
628 - each process now have to call gras_{init,exit}. One day, their log
629 settings will be separated
630 - Continue the code factorisation between SG, RL and common in Transport.
632 2004-07-04 Martin Quinson
634 - Redistribution between SG and RL.
635 We wanna have to accept in SG, so move accepted related parts of RL in
636 the common part. (more precisely, the dynar of all known sockets is no
637 more a static in transport.c, but part of the process_data)
639 [gras_stub_generator]
640 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
642 2004-07-03 Martin Quinson
643 - Create a new log channel tbx containing dict, set, log, dynar (to shut
644 them all up in one shot)
646 - Fix the ugly case of reference to dynamic array.
647 - New (semi-public) function gras_datadesc_size to allow the messaging
648 layer to malloc the needed space for the buffer.
650 - gras_socket_close now expect the socket to close (and not its address to
651 put NULL in it after it). This is because the socket passed to handlers
652 is one of their argument (=> not writable).
654 - propagate the interface cleanup from last week in datadesc, ie remove a
655 superfluous level of indirection. User pass adress of variable
656 containing data (both when sending and receiving), and not of a variable
657 being a pointer to the data. Let's say that I like it better ;)
658 The price for that is constructs like "int msg=*(int*)payload" in
659 handlers, but it's a fine price, IMHO.
661 - Let it work in RL (yuhu)
663 2004-06-21 Martin Quinson
665 - porting SG plugin and SG select to new standards (works almost).
666 - plug memleaks and fix bugs around.
669 - cleanup the prototype of data recv and force users to specify when they
670 want to handle references to objects. Test case working even for cycles.
671 - plug memleaks. Valgrind is perfectly ok with this.
673 2004-06-12 Martin Quinson
675 - cleanup the separation between plugin and main code in plugin creation
677 2004-06-11 Martin Quinson
679 - Reput hook for raw sockets, needed for BW experiments
680 - kill a few lines of dead code
681 [Data description] Interface cleanup
682 - gras_datadesc_by_name returns the searched type or NULL.
683 That way, no variable is needed to use a type desc once, which makes
685 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
686 two parameters were strings (field name, type name), leading to
688 [Dicos] Interface cleanup
689 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
690 This is consistant with the dynar API.
692 2004-04-21 Martin Quinson
694 - Porting to new standards.
697 There is no bag anymore, no need to take extra provision to mask the
698 pointers behind "ID".
699 Better splitup of functions between files create/exchange/convert.
700 This is still a bit artificial since convert and receive are so
701 interleaved, but anyway.
703 - add a queued message list to procdata (the ones not matching criteria
705 - factorize some more code between SG and RL wrt procdata
707 - use gras_exit in example to track memleaks
708 - get rid of gs_example now that GS is properly integrated into gras
709 - update run_test to integrate the lastest tests (datadesc)
711 - rename WARNINGn macros to WARNn since it prooved error-prone
713 2004-04-19 Martin Quinson
715 - register init/exit functions within gras module mechanism
716 - send/receive function.
717 Convertion is not implemented, but short-cutted if not needed.
718 struct/array elements are sent one by one (instead of block-wise), but
719 nobody really cares (yet). Get a prototype before optimizing.
720 - tests (using a file socket) for DD send/receive on:
721 - base types: int, float
722 - array: fixed size, string (ie ref to dynamic string)
723 - structure: homogeneous, heterogeneous
724 - chained list, graph with cycle
725 Believe it or not, valgrind is not too unhappy with the results. The
726 cycle happily segfaults, but the others are ok. And I'm sick of pointers
730 - Bugfix when using a filename explicitely (instead of '-')
732 2004-04-09 Martin Quinson
734 - factorize more code between RL and SG in socket creation
735 - Complete the implementation and tests of:
737 o file (only in RL, and mainly for debugging)
739 I lost 3 days to design a portable address resolver, and then decided
740 that the prototype mainly have to run on my box.
741 Addressing portability too early may be like optimizing too early :-/
743 - use gras_init in the Tests instead of the crappy parse_log_opt
744 (the latter function is removed)
745 [Conditional execution]
746 - New functions: gras_if_RL/gras_if_SG (basic support for this)
747 [Code reorganisation]
748 - Get rid of libgrasutils.a since it makes more trouble than it solves.
749 Build examples against the RL library, since there is no way to disable
750 its creation for now.