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]
20 * Add addtionnal checkings on channel values in communicating functions.
23 Documentation improvements:
24 * xbt_log module unmercifully reworked [MQ]
28 SimGrid (3.00) stable; urgency=low
31 * New! Give the possibility to hijack the surf parser and thus bypass
32 MSG_create_environment and MSG_launch_application. Have a look at
33 examples/msg/msg_test_surfxml_bypassed.c to see how it can be done.
35 -- Arnaud Legrand <simgrid-devel@lists.gforge.inria.fr> Sat, 20 Aug 2005 23:25:25 -0700
37 SimGrid (2.96) unstable; urgency=low
42 * New! Exception handling with setjmp or such (code from OSSP ex) [MQ]
43 This deprecates the xbt_error_t mecanisms.
44 It modifies (simplifies) all XBT and GRAS API.
45 MSG API keeps unchanged (exceptions raised by XBT are catched from
46 within MSG and masked with existing error handling facilities)
49 * New! Add a FATPIPE model. [AL]
50 * New! Add a parallel task model. [AL]
51 * New! Add automatically a loopback interface (in the default
52 network model) if none was precised.
55 * Bugfix: MSG_process_resume now works with the current running process.
57 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
58 * Modification of MSG_task_get_compute_duration. Once a task has been
59 processed, the value returned by this function is now equal to 0. [AL]
60 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
61 MSG_error_t MSG_task_cancel(m_task_t task). Add a state
62 (MSG_TASK_CANCELLED) to MSG_error_t corresponding to the cancelation
63 of a m_task. For now, MSG_task_cancel only works with computation
65 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
66 of the processor (in Mflop/s) regardless of the current load on the
68 * API Change: use proper naming convention for MSG_getClock and
69 MSG_process_isSuspended: MSG_get_clock and MSG_process_is_suspended.
71 * New! Add void MSG_task_set_priority(m_task_t task, double priority).
72 This function changes the priority of a computation task. This priority
73 doesn't affect the transfer rate. A priority of 2 will make a task
74 receive two times more cpu power than the other ones. This function
75 has been added to suit the needs of Nguyen The Loc and hasn't been that
76 much tested yet. So if it fails, please report it and send me your code.
78 * API Change: removed all functions and types that were marked "deprecated"
79 since many months. Renamed MSG_global_init_args to MSG_global_init.
81 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Mon, 8 Aug 2005 17:58:47 -0700
83 SimGrid (2.95) unstable; urgency=low
86 * Steal some nice code to GNU pth to fix context detection and usage [AL]
87 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
88 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
91 * Bugfix: Allow absolute paths for platform description files [MQ]
92 * Bugfix: do free the variables after use. Leads to drastic performance
94 * Implement max_duration (ie, timeouts) on resources [AL]
97 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
99 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
100 the needed support (timeouts) [AL]
103 * Implement measurement sockets. You can now get the bandwidth between two
104 hosts thanks to AMOK (see below). [MQ]
105 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
106 dynar over the network (yeah) [MQ]
107 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
108 * simple/static token-ring example. [Alexandre Colucci and MQ]
109 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
113 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
114 type parsing. "short" and "long" modifiers were also ignored; other
115 modifier (such as reference level) are still ignored. [MQ]
116 * Embeed the buffer size within the buffer itself on SG. [MQ]
117 That way, send() are atomic and cannot get intermixed anymore (at least
118 the ones which are less than 100k; bigger messages still have the issue)
119 * Array size pushed by the field, not by the field type (or each
120 and every long int will push stuff to the cbps) [MQ]
121 * use select() to sleep since it allows to portably sleep less than one
124 GRAS (minor cleanups)
125 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
128 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
129 * type category 'ignored' killed as it was never used and were difficult
131 * whether a type can cycle or not is now a flag, leaving room for more
132 flags (as "ignored", if we feel the need one day ;)
133 * Rename raw sockets to measurement sockets since "raw" has another
134 meaning in networking community.
137 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
138 and offers features not belonging to GRAS but that most applications
139 need. One day, it may be a set of plugins loadable at runtime.
140 * New module: bandwidth
141 bandwidth measurement between arbitrary nodes running this module. [MQ]
143 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
145 SimGrid (2.94) unstable; urgency=low
147 The first beta release of SimGrid 3 !
151 * Update the main page and the FAQ. Adding references to gforge.
154 * Add a gras_os_getpid function.
157 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
158 * Extend the logs so that they also print PID, hostname, date, ... if
160 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
161 and kill the old version which were in testsuite/
162 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
166 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
167 for users. Instead, each programming environment (GRAS, MSG,...) use it
168 when needed to provide such a feature to users.
169 Don't shortcut the mecanism or you will also shortcut the virtualization
170 you need on the simulator.
174 * Cleanups in configury with regard to compile optimization/warning flags.
175 Also add -fno-loop-optimize to any powerpc since it's the optimization
176 killing gcc (< 3.4.0).
177 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
178 needed by MSG examples complications
179 * Borrow configury beautifications from PHP
182 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
183 warning (thanks loris for stumbling into this one).
184 * Bugfix: stop loading private headers (gras_config.h) from the public
188 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
189 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
190 RL would imply a DNS resolver, which is *hard* to do in a portable way
191 (and therefore delayed).
192 * Implement a real timer mecanism and use it in timing macros. This allows
193 to avoid rounding errors and get a 0.000005 sec precision in timing
194 macros. While I was at it, various cleanups:
195 - allow to declare more than one timed section per file (fix a stupid bug)
196 - move some private declaration to the right place
197 - merge conditional execution and timing macros into emulation module
198 - document the module
199 - make sure the module cleanups its mess on gras_exit
200 * Documentation improvements:
201 - (new) how to compile applications using GRAS
202 - (new) emulation support (timing macros)
204 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
206 SimGrid (2.93) unstable; urgency=low
208 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
211 - Use Paje properly where used. Still to be sanitized properly.
212 - Portability fix: Add an implementation of the contexts using pthread
216 - Add xbt_procname(): returns the name of the current process.
217 Use it to show the current process's name in all logging.
219 - fix detection of older flex version and the reaction, since we do
220 depend on modern ones (we use lex_destroy)
221 - Better separation of SG and RL in the libs: remove all simulation code
222 from libgras. As a result, this lib is now only 200k when stripped.
223 Some of the xbt modules may also be duplicated (two sets and such) and
224 should be cleaned/killed before SG3.
225 - Insist on using xlC on AIX because of weird problems involving gcc there.
226 - Cleanup the make remote stuff. This is now done by scripts
227 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
228 mainly for our private use, but we're working on changing them to user
231 - Bugfix: flush the socket on close only if there is some *output*.
232 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
234 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
237 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
239 SimGrid (2.92) unstable; urgency=low
241 Alpha 3 on the path to SimGrid 3
245 - New! First try of benchmarking macros.
246 - New! First try so that gras_stub_generator generate deployment and
247 remote compilation helpers.
249 - Bugfix: Initialization fix in msg_test.
253 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
255 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
257 - API Change: gras_os_sleep to take the amount of seconds as a double.
258 Accepting an int was error prone since it was the only location where
259 seconds were coded as such. It leaded to damn rounding errors.
260 - Bugfix: Hard to belive that timers ever worked before this.
262 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
264 SimGrid (2.91) unstable; urgency=low
266 Alpha 2 on the path to SimGrid 3
270 - Bug fix in the lmm_solver.
272 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
273 through the function MSG_paje_output.
274 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
275 - It is possible to bound the rate of a communication in MSG with
276 MSG_task_put_bounded() (was already in the previous version but I had forgotten
277 to write it in the changelog).
278 - Bug fix to let GRAS run on top of MSG until we move it directly on top
283 - Various cleanups to the autotools stuff
284 - Begin to move Gras examples to examples/gras/
285 - Let make distcheck work again (yeah!)
287 - documentation overhauled using doxygen.
288 gtk-doc-tools is dead in SimGrid now.
289 - Automatically extract all existing logging categories, and add the list
290 to the documentation (long standing one, to say the less)
292 - Cleanup the known architecture table. Reorder the entries to group what
293 should be, and use a more consistent naming scheme.
294 (some of the test dataset are still to be regenerated)
295 - New! Allow library to register globals on each process just as userdata
297 This is implemented using a xbt_dict and not a xbt_set, so we loose the
298 lookup time (for now).
299 Use it in msg and trp.
300 This cleans a lot the internals and helps enforcing privacy of the
301 headers between the gras components.
302 - New! Add a timer mechanism, not unlike cron(8) and at(1).
303 - Bugfix: gras_os_time was delirious in RL.
304 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
306 - Reenable GRAS now that it works.
308 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
310 SimGrid (2.90) unstable; urgency=low
312 Alpha 1 on the path to SimGrid 3
314 * It is a long time since the last release of SimGrid. I'm sorry about
315 that but as I had told you, I was rewriting a lot of things. I apologize
316 to those who had been reporting bugs to me and that I had not answered.
317 If your bug is still in the new version, please tell me. Here is a
318 summary of the main changes.
320 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
321 lead by Martin Quinson. As a consequence SimGrid gains a lot in
322 portability, speed, and a lot more but you'll figure it out later.
323 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
324 I wanted to release the new MSG as soon as possible and I have
325 broken GRAS, which is the reason why, for now, only MSG is fully
326 functional. A laconic description of these projects is available
327 in the documentation.
329 * REVOLUTION 2: I have removed SG and I am now using a new simulation
330 kernel optimized for our needs (called SURF but only the developers
331 should use it). Hence, MSG is now roughly 30 times faster and I think
332 that by rewriting a little bit MSG, I could event speed it up a little
333 bit more. Beside the gain in speed, it is also much easier to encode a
334 new platform model with SURF than it was with SG. More to come...
336 * REVOLUTION 3: I have tried to change a little as possible the API of
337 MSG but a few things really had to disappear. The main differences
338 with the previous version are :
339 1) no more m_links_t and the corresponding functions. Platforms are
340 directly read from a XML description and cannot be hard-coded
341 anymore. The same format is used for application deployment
342 description. The new format is described in the documentation.
343 Have a look in tools/platform_generation. There is a tiny script
344 that converts from the old platform format to the new one. Concerning
345 the application deployment format, parsing the old one is tricky.
346 I think most of you should however be able to convert your files. If
347 it is really an issue, I can write a C code that does the conversion.
349 2) the toolbox tbx does not exist anymore. We now have a library
350 with much more data-structures but without the hash-tables (we have
351 dictionaries that are much faster).
353 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
355 *****************************************************************************
356 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
357 * don't feel like converting the oldies. (Mt) *
358 *****************************************************************************
361 Version 2.90: "the long awaited one"
362 - Finished rewriting and debugging MSG. Rewrote the documentation.
363 - disable GRAS for now since it needs to be ported to the newest SG
366 - Finish the port to windows (using mingw32 for cross-compile)
369 - Main loop and datastructures of SURF. A cpu resource object is
370 functional. Surf can thus be used to create cpu's with variable
371 performance on which you can execute some actions.
373 2004-11-15 Martin Quinson
374 - Port to ARM. Simply added the alignment and size descriptions. Should
375 work, but the ARM machines are so slow that I didn't had the opportunity
376 to 'make check' over there yet.
378 2004-11-15 Arnaud Legrand
379 - Trace manager now written. It uses a heap structure and is therefore
380 expected to be efficient. It may however be speeded up (particularly
381 when many events occur at the same date) by using red and black
382 trees. One day maybe...
383 - Max-min linear system solver written. It uses a sparse matrix
384 structure taking advantage of its expected use. Most operations are
385 O(1) and free/calloc are called as few as possible. The computation of
386 the minimum could however be improved by using a red and black tree
389 2004-11-03 Arnaud Legrand
390 - Rename every gras_* function that was in xbt/ to its xbt_
392 - Add a heap and a doubly-linked list to xbt
393 - Added a dichotomy to the dictionaries. make check works as well before
394 so I assume that the patch is correct. I do not know however if things
395 run effectively faster than before now. :)
397 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
398 SimGrid, and the version number is bumped to 2.x
400 2004-10-29 Martin Quinson
401 - Introduction of the remote errors.
402 They are the result of a RMI/RPC on the remote machine.
403 ErrCodes being scalar values, you can't get the host on which those
404 errors did happen. Extending the error mechanism as in Gnome is possible.
405 No idea yet whether it is a good idea.
407 2004-10-28 Martin Quinson
408 - Interface revolution: the Starred Structure Eradication.
409 I used to do typedef struct {} toto_t; and then handle *toto_t.
410 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
411 - ???_t is a valid type (builded with typedef)
412 - s_toto_t is a structure (access to fields with .)
413 - s_toto is a structure needing 'struct' keyword to be used
414 - e_toto_t is an enum
415 - toto_t is an 'object' (struct*)
417 typedef struct s_toto {} s_toto_t, *toto_t;
418 typedef enum {} e_toto_t;
419 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
420 s_toto_t) is private.
422 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
423 so that it can set the variable to NULL. It was so for dicts and sets,
424 it changed for dynars.
426 - Fix a bunch of memleaks in dict_remove
427 - Fix a bug in sg/server_socket opening: it failed all the time.
429 2004-10-07 Martin Quinson
430 - Speed up dynar lookup operation a bit.
432 gras_dynar_get is dead.
434 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
435 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
436 (which gives you the address of the stored data).
438 gras_dynar_get_as is an helpful macro which allows you to retrieve a
439 copy of the data using an affectation to do the job and not a memcpy.
441 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
442 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
444 It does not really speedup the dynar test because they are
445 setting elements all the time (and look them seldom). But the dict does
446 far more lookup than setting.
448 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
450 2004-10-05 Martin Quinson
451 - Allow to (en/dis)able the cycle detection at run time.
453 Whether we should check for cycle or not is now a property of each
454 datatype. When you think there may be some cycle, use datadesc_cycle_set.
455 datadesc_cycle_unset allow to remove this property when previously set.
457 Note that the cycle detection is off by default since it impacts the
458 performance. Watch the data you feed GRAS with ;)
460 This property is hereditary. Any element embedded in a structure having it
461 set have it set for the time of this data exchange.
463 You should set it both on sender and receiver side. If you don't set it on
464 sender side, it will enter an endless loop. If you forget on receiver
465 side, the cycles won't be recreated after communication.
467 - Header reorganization.
468 Kill gras_private.h, each submodule must load the headers it needs.
470 2004-10-04 Martin Quinson
471 - Interface revolution: do not try to survive to malloc failure.
473 Now, gras_malloc and friends call gras_abort() on failure.
474 As a conclusion, malloc_error is not a valid error anymore, and all
475 functions for which it was the only gras_error_t return value are
476 changed. They now return void, or there result directly.
477 This simplify the API a lot.
479 2004-09-29 Martin Quinson
480 - Re-enable raw sockets.
481 Created by gras_socket_{client,server}_ext;
482 Used with gras_raw_{send,recv}
485 It should allow to kill the last bits of gras first version soon.
487 This is not completely satisfactory yet (duplicate code with
488 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
491 - Simplify transport plugin (internal) interface by not passing any
492 argument to _server and _client, but embedding them in the socket
495 2004-09-28 Martin Quinson
496 - Finish the port to AIX.
497 autoconf was my problem (segfault within the malloc replacement
498 function. No idea why)
500 2004-09-16 Martin Quinson
501 - Fix some size_t madness on 64bit architectures.
503 2004-09-08 Martin Quinson
504 - Reduce the number of system headers loaded, overload some more system
505 calls (such as malloc to cast the result of the system one, and work
507 - Fix and reintroduce the config support
509 2004-09-07 Martin Quinson
510 - Source code reorganization to allow Arnaud to surf all over there.
511 - Allow to document the logging categories.
512 - Remove all uppercase from logging categories and useless cleanup in names.
514 2004-08-18 Martin Quinson
515 Version 0.6.2 (protocol not changed; API changed)
516 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
517 gras_error_t), and NULL when not found. Functions expecting a msgtype
518 as argument (msg_wait; msg_send) deal with NULL argument by providing a
519 hopefully usefull message.
520 - Portability to prehistoric sparcs again
522 2004-08-17 Martin Quinson
523 Version 0.6.1 (protocol not changed; ABI not changed)
524 - prealloc some buffers to speed things up
526 2004-08-11 Martin Quinson
527 Version 0.6 (protocol not changed; ABI expended)
528 - The parsing macro can deal with the references, provided that you add
529 the relevant annotations (using GRAS_ANNOTE(size,field_name))
531 2004-08-09 Martin Quinson
532 Version 0.5 (protocol not changed; ABI changed)
533 - Allow to off turn the cycle detection code in data exchange at
534 compilation time. It should be at run time, but I'm short of time (and
535 the config stuff is still broken). That way, we keep dict out of the
536 critical path, which is good because the performance is poor:
537 - search not dichotomial yet
538 - dynar give no way to access their content and memcpy everytime
539 - In composed data description (struct, ref and so on), stop foolness of
540 keeping the subtype's ID, but store the type itself. This keeps sets out
541 of the critical path, which is good since they rely on dynar and
542 dictionnaries. The only loose of that is that we cannot detect the
543 redeclaration of a structure/union with another content (but I'm not sure
544 the code detected well this error before anyway). We still can detect
545 the redefinition discrepancy for the other types.
546 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
547 it breaks the code because of type-punning used all over the place).
548 This breaks on all non-gcc architectures (for now).
550 All those changes (plus the buffer of last time) allow me to gain 2 order
551 of magnitude on cruel tests consisting of 800000 array of integers on two
552 level of a hierarchical structure (200 secondes -> 4 secondes)
555 - the selector of reference must now return the type it points to, not
558 2004-08-06 Martin Quinson
559 Version 0.4 (protocol changed; ABI not changed)
560 - Allow to pass --gras-log argument to processes in simulation mode. Really.
561 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
562 - Implement a buffer transport, and use it by default (it relies on tcp in
563 real life and on sg in simulation).
564 That's a bit hackish since I had a new field to the structure to store
565 its data without interfering with the subtype ones. Inheritance
566 is tricky in C. And that's a kind of reverse inheritance with one class
567 derivating two classes. Or maybe a game with java interfaces. Anyway,
568 that's damn hard in C (at least).
569 Moreover, I got tired while trying to ensure plugin separation and
570 genericity in SG mode. MSG wants me to do weird things, so let's go for
571 cruel hacks (temporarily of course ;).
572 See comment in transport_private.h:71
573 - do not include all the _interface headers in private but in the files
574 which really need them (to cut the compilation time when they are
577 2004-07-26 Martin Quinson
578 Version 0.3 (protocol not changed; ABI changed)
579 - Major overhault of the datadesc interface to simplify it:
580 - shorted the function names:
581 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
582 - add a trivial way to push/pop integers into the cbps without malloc.
583 This allows to make really generic sub_type description, which simply
584 pop their size of the stack.
585 - add a function gras_datadesc_ref_pop_arr() which does what users want
586 most of the time: Declare a dynamic array (which pops its size of the
587 stack) and declare a reference to it. Poor name, but anyway.
588 - kill the post-send callback, add a post-receive one
590 2004-07-23 Martin Quinson
591 Version 0.2 (protocol changed; ABI changed)
592 - add some testing for cpbs in the test cases, and fix some more bugs.
593 This invalidate again the little64 data file, since I cannot regenerate
595 - remove an awfull optimization in the logging stuff, allowing me to:
596 - understand it again
597 - learn gcc how to check that the argument match the provided format
598 - fix all errors revealed by gcc after that
599 - internal keys of dict are not \0 terminated. Deal with it properly in
600 loggings instead of segfaulting when the user want to see the logs :-/
602 2004-07-22 Martin Quinson
603 - Fix some stupid bug preventing cbps (callback postit) from working
605 2004-07-21 Martin Quinson
606 - Some documentation cleanups
607 - remove the useless last argument of msgtype_declare
608 - rename the Virtu functions to fit into the 'os' namespace
609 - move headers src/include -> src/include/gras/ and stop fooling with
610 gras -> . symbolic link
611 - make distcheck is now successful
613 2004-07-19 Martin Quinson
615 - Build shared library also
616 - Install html doc to the right location
617 - stop removing maintainer files in make clean
618 - build tests only on make check
620 2004-07-13 Martin Quinson
622 - No major issue in previous version => change versionning schema
623 - Re-enable little64 convertion test now that Abdou kindly regenerated the
624 corresponding dataset.
626 2004-07-11 Martin Quinson
628 - Get it working with any kind of structure (we can compute the padding
629 bytes remotely for all the architectures I have access to)
630 - Implement the structure parsing macro (still not quite robust/complete)
631 - Improvement to the remote testing toysuite
633 2004-07-10 Martin Quinson
635 - get ride of a bunch of deprecated macros
636 - actually run the test for two-compliment, not only compile it :-/
637 - test whether the structures get packed (and bail out if yes. Damn.
638 Alignment is a serious matter)
639 - test whether the structures get compacted (but respecting the alignment
640 constraints of each types)
641 - test whether the array fields of structures can straddle alignment boundaries
643 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
644 (took me 2 hours to find that bug, looking at the wrong place)
645 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
646 before sending/receiving and are used to compute the offsets of fields
647 - Given that padding size depend even on compiler options, keep track of
648 alignment and aligned_size only for the current architecture. Not a big
649 deal since we send structure fields one after the other (seems
651 - Add the datastructure used for IEEE paper by the PBIO guys to the test
652 program, let it work on linux/gcc/little32. portability todo.
654 2004-07-08 Martin Quinson
655 - import and improve remote compilation support from FAST
656 - make sure make check works on half a dozen of machines out there
658 2004-07-07 Martin Quinson
659 Let's say it's version 0.0.3 ;)
660 - Implement conversions (yuhu!)
661 - Let it work on solaris (beside conversion, of course)
662 - Stupid me, using rand() to generate the conversion datatests in not wise.
664 2004-07-06 Martin Quinson
665 - Let make dist work, since I'm gonna need it to compile on remote hosts
666 - Let Tests/datadesc_usage write the architecture on which the file was
667 generated as first byte.
668 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
669 architecture descriptions.
670 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
671 on those architectures.
672 - Optimization: send/recv array of scalar in one shoot
674 2004-07-05 Martin Quinson
675 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
677 - Plug a whole bunch of memleaks
678 - each process now have to call gras_{init,exit}. One day, their log
679 settings will be separated
680 - Continue the code factorisation between SG, RL and common in Transport.
682 2004-07-04 Martin Quinson
684 - Redistribution between SG and RL.
685 We wanna have to accept in SG, so move accepted related parts of RL in
686 the common part. (more precisely, the dynar of all known sockets is no
687 more a static in transport.c, but part of the process_data)
689 [gras_stub_generator]
690 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
692 2004-07-03 Martin Quinson
693 - Create a new log channel tbx containing dict, set, log, dynar (to shut
694 them all up in one shot)
696 - Fix the ugly case of reference to dynamic array.
697 - New (semi-public) function gras_datadesc_size to allow the messaging
698 layer to malloc the needed space for the buffer.
700 - gras_socket_close now expect the socket to close (and not its address to
701 put NULL in it after it). This is because the socket passed to handlers
702 is one of their argument (=> not writable).
704 - propagate the interface cleanup from last week in datadesc, ie remove a
705 superfluous level of indirection. User pass adress of variable
706 containing data (both when sending and receiving), and not of a variable
707 being a pointer to the data. Let's say that I like it better ;)
708 The price for that is constructs like "int msg=*(int*)payload" in
709 handlers, but it's a fine price, IMHO.
711 - Let it work in RL (yuhu)
713 2004-06-21 Martin Quinson
715 - porting SG plugin and SG select to new standards (works almost).
716 - plug memleaks and fix bugs around.
719 - cleanup the prototype of data recv and force users to specify when they
720 want to handle references to objects. Test case working even for cycles.
721 - plug memleaks. Valgrind is perfectly ok with this.
723 2004-06-12 Martin Quinson
725 - cleanup the separation between plugin and main code in plugin creation
727 2004-06-11 Martin Quinson
729 - Reput hook for raw sockets, needed for BW experiments
730 - kill a few lines of dead code
731 [Data description] Interface cleanup
732 - gras_datadesc_by_name returns the searched type or NULL.
733 That way, no variable is needed to use a type desc once, which makes
735 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
736 two parameters were strings (field name, type name), leading to
738 [Dicos] Interface cleanup
739 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
740 This is consistant with the dynar API.
742 2004-04-21 Martin Quinson
744 - Porting to new standards.
747 There is no bag anymore, no need to take extra provision to mask the
748 pointers behind "ID".
749 Better splitup of functions between files create/exchange/convert.
750 This is still a bit artificial since convert and receive are so
751 interleaved, but anyway.
753 - add a queued message list to procdata (the ones not matching criteria
755 - factorize some more code between SG and RL wrt procdata
757 - use gras_exit in example to track memleaks
758 - get rid of gs_example now that GS is properly integrated into gras
759 - update run_test to integrate the lastest tests (datadesc)
761 - rename WARNINGn macros to WARNn since it prooved error-prone
763 2004-04-19 Martin Quinson
765 - register init/exit functions within gras module mechanism
766 - send/receive function.
767 Convertion is not implemented, but short-cutted if not needed.
768 struct/array elements are sent one by one (instead of block-wise), but
769 nobody really cares (yet). Get a prototype before optimizing.
770 - tests (using a file socket) for DD send/receive on:
771 - base types: int, float
772 - array: fixed size, string (ie ref to dynamic string)
773 - structure: homogeneous, heterogeneous
774 - chained list, graph with cycle
775 Believe it or not, valgrind is not too unhappy with the results. The
776 cycle happily segfaults, but the others are ok. And I'm sick of pointers
780 - Bugfix when using a filename explicitely (instead of '-')
782 2004-04-09 Martin Quinson
784 - factorize more code between RL and SG in socket creation
785 - Complete the implementation and tests of:
787 o file (only in RL, and mainly for debugging)
789 I lost 3 days to design a portable address resolver, and then decided
790 that the prototype mainly have to run on my box.
791 Addressing portability too early may be like optimizing too early :-/
793 - use gras_init in the Tests instead of the crappy parse_log_opt
794 (the latter function is removed)
795 [Conditional execution]
796 - New functions: gras_if_RL/gras_if_SG (basic support for this)
797 [Code reorganisation]
798 - Get rid of libgrasutils.a since it makes more trouble than it solves.
799 Build examples against the RL library, since there is no way to disable
800 its creation for now.