1 SimGrid (2.96) unstable; urgency=low
4 * Bugfix: MSG_process_resume now works with the current running process.
8 SimGrid (2.95) unstable; urgency=low
11 * Steal some nice code to GNU pth to fix context detection and usage [AL]
12 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
13 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
16 * Bugfix: Allow absolute paths for platform description files [MQ]
17 * Bugfix: do free the variables after use. Leads to drastic performance
19 * Implement max_duration (ie, timeouts) on resources [AL]
22 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
24 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
25 the needed support (timeouts) [AL]
28 * Implement measurement sockets. You can now get the bandwidth between two
29 hosts thanks to AMOK (see below). [MQ]
30 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
31 dynar over the network (yeah) [MQ]
32 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
33 * simple/static token-ring example. [Alexandre Colucci and MQ]
34 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
38 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
39 type parsing. "short" and "long" modifiers were also ignored; other
40 modifier (such as reference level) are still ignored. [MQ]
41 * Embeed the buffer size within the buffer itself on SG. [MQ]
42 That way, send() are atomic and cannot get intermixed anymore (at least
43 the ones which are less than 100k; bigger messages still have the issue)
44 * Array size pushed by the field, not by the field type (or each
45 and every long int will push stuff to the cbps) [MQ]
46 * use select() to sleep since it allows to portably sleep less than one
50 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
53 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
54 * type category 'ignored' killed as it was never used and were difficult
56 * whether a type can cycle or not is now a flag, leaving room for more
57 flags (as "ignored", if we feel the need one day ;)
58 * Rename raw sockets to measurement sockets since "raw" has another
59 meaning in networking community.
62 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
63 and offers features not belonging to GRAS but that most applications
64 need. One day, it may be a set of plugins loadable at runtime.
65 * New module: bandwidth
66 bandwidth measurement between arbitrary nodes running this module. [MQ]
68 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
70 SimGrid (2.94) unstable; urgency=low
72 The first beta release of SimGrid 3 !
76 * Update the main page and the FAQ. Adding references to gforge.
79 * Add a gras_os_getpid function.
82 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
83 * Extend the logs so that they also print PID, hostname, date, ... if
85 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
86 and kill the old version which were in testsuite/
87 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
91 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
92 for users. Instead, each programming environment (GRAS, MSG,...) use it
93 when needed to provide such a feature to users.
94 Don't shortcut the mecanism or you will also shortcut the virtualization
95 you need on the simulator.
99 * Cleanups in configury with regard to compile optimization/warning flags.
100 Also add -fno-loop-optimize to any powerpc since it's the optimization
101 killing gcc (< 3.4.0).
102 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
103 needed by MSG examples complications
104 * Borrow configury beautifications from PHP
107 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
108 warning (thanks loris for stumbling into this one).
109 * Bugfix: stop loading private headers (gras_config.h) from the public
113 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
114 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
115 RL would imply a DNS resolver, which is *hard* to do in a portable way
116 (and therefore delayed).
117 * Implement a real timer mecanism and use it in timing macros. This allows
118 to avoid rounding errors and get a 0.000005 sec precision in timing
119 macros. While I was at it, various cleanups:
120 - allow to declare more than one timed section per file (fix a stupid bug)
121 - move some private declaration to the right place
122 - merge conditional execution and timing macros into emulation module
123 - document the module
124 - make sure the module cleanups its mess on gras_exit
125 * Documentation improvements:
126 - (new) how to compile applications using GRAS
127 - (new) emulation support (timing macros)
129 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
131 SimGrid (2.93) unstable; urgency=low
133 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
136 - Use Paje properly where used. Still to be sanitized properly.
137 - Portability fix: Add an implementation of the contexts using pthread
141 - Add xbt_procname(): returns the name of the current process.
142 Use it to show the current process's name in all logging.
144 - fix detection of older flex version and the reaction, since we do
145 depend on modern ones (we use lex_destroy)
146 - Better separation of SG and RL in the libs: remove all simulation code
147 from libgras. As a result, this lib is now only 200k when stripped.
148 Some of the xbt modules may also be duplicated (two sets and such) and
149 should be cleaned/killed before SG3.
150 - Insist on using xlC on AIX because of weird problems involving gcc there.
151 - Cleanup the make remote stuff. This is now done by scripts
152 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
153 mainly for our private use, but we're working on changing them to user
156 - Bugfix: flush the socket on close only if there is some *output*.
157 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
159 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
162 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
164 SimGrid (2.92) unstable; urgency=low
166 Alpha 3 on the path to SimGrid 3
170 - New! First try of benchmarking macros.
171 - New! First try so that gras_stub_generator generate deployment and
172 remote compilation helpers.
174 - Bugfix: Initialization fix in msg_test.
178 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
180 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
182 - API Change: gras_os_sleep to take the amount of seconds as a double.
183 Accepting an int was error prone since it was the only location where
184 seconds were coded as such. It leaded to damn rounding errors.
185 - Bugfix: Hard to belive that timers ever worked before this.
187 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
189 SimGrid (2.91) unstable; urgency=low
191 Alpha 2 on the path to SimGrid 3
195 - Bug fix in the lmm_solver.
197 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
198 through the function MSG_paje_output.
199 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
200 - It is possible to bound the rate of a communication in MSG with
201 MSG_task_put_bounded() (was already in the previous version but I had forgotten
202 to write it in the changelog).
203 - Bug fix to let GRAS run on top of MSG until we move it directly on top
208 - Various cleanups to the autotools stuff
209 - Begin to move Gras examples to examples/gras/
210 - Let make distcheck work again (yeah!)
212 - documentation overhauled using doxygen.
213 gtk-doc-tools is dead in SimGrid now.
214 - Automatically extract all existing logging categories, and add the list
215 to the documentation (long standing one, to say the less)
217 - Cleanup the known architecture table. Reorder the entries to group what
218 should be, and use a more consistent naming scheme.
219 (some of the test dataset are still to be regenerated)
220 - New! Allow library to register globals on each process just as userdata
222 This is implemented using a xbt_dict and not a xbt_set, so we loose the
223 lookup time (for now).
224 Use it in msg and trp.
225 This cleans a lot the internals and helps enforcing privacy of the
226 headers between the gras components.
227 - New! Add a timer mechanism, not unlike cron(8) and at(1).
228 - Bugfix: gras_os_time was delirious in RL.
229 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
231 - Reenable GRAS now that it works.
233 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
235 SimGrid (2.90) unstable; urgency=low
237 Alpha 1 on the path to SimGrid 3
239 * It is a long time since the last release of SimGrid. I'm sorry about
240 that but as I had told you, I was rewriting a lot of things. I apologize
241 to those who had been reporting bugs to me and that I had not answered.
242 If your bug is still in the new version, please tell me. Here is a
243 summary of the main changes.
245 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
246 lead by Martin Quinson. As a consequence SimGrid gains a lot in
247 portability, speed, and a lot more but you'll figure it out later.
248 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
249 I wanted to release the new MSG as soon as possible and I have
250 broken GRAS, which is the reason why, for now, only MSG is fully
251 functional. A laconic description of these projects is available
252 in the documentation.
254 * REVOLUTION 2: I have removed SG and I am now using a new simulation
255 kernel optimized for our needs (called SURF but only the developers
256 should use it). Hence, MSG is now roughly 30 times faster and I think
257 that by rewriting a little bit MSG, I could event speed it up a little
258 bit more. Beside the gain in speed, it is also much easier to encode a
259 new platform model with SURF than it was with SG. More to come...
261 * REVOLUTION 3: I have tried to change a little as possible the API of
262 MSG but a few things really had to disappear. The main differences
263 with the previous version are :
264 1) no more m_links_t and the corresponding functions. Platforms are
265 directly read from a XML description and cannot be hard-coded
266 anymore. The same format is used for application deployment
267 description. The new format is described in the documentation.
268 Have a look in tools/platform_generation. There is a tiny script
269 that converts from the old platform format to the new one. Concerning
270 the application deployment format, parsing the old one is tricky.
271 I think most of you should however be able to convert your files. If
272 it is really an issue, I can write a C code that does the conversion.
274 2) the toolbox tbx does not exist anymore. We now have a library
275 with much more data-structures but without the hash-tables (we have
276 dictionaries that are much faster).
278 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
280 *****************************************************************************
281 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
282 * don't feel like converting the oldies. (Mt) *
283 *****************************************************************************
286 Version 2.90: "the long awaited one"
287 - Finished rewriting and debugging MSG. Rewrote the documentation.
288 - disable GRAS for now since it needs to be ported to the newest SG
291 - Finish the port to windows (using mingw32 for cross-compile)
294 - Main loop and datastructures of SURF. A cpu resource object is
295 functional. Surf can thus be used to create cpu's with variable
296 performance on which you can execute some actions.
298 2004-11-15 Martin Quinson
299 - Port to ARM. Simply added the alignment and size descriptions. Should
300 work, but the ARM machines are so slow that I didn't had the opportunity
301 to 'make check' over there yet.
303 2004-11-15 Arnaud Legrand
304 - Trace manager now written. It uses a heap structure and is therefore
305 expected to be efficient. It may however be speeded up (particularly
306 when many events occur at the same date) by using red and black
307 trees. One day maybe...
308 - Max-min linear system solver written. It uses a sparse matrix
309 structure taking advantage of its expected use. Most operations are
310 O(1) and free/calloc are called as few as possible. The computation of
311 the minimum could however be improved by using a red and black tree
314 2004-11-03 Arnaud Legrand
315 - Rename every gras_* function that was in xbt/ to its xbt_
317 - Add a heap and a doubly-linked list to xbt
318 - Added a dichotomy to the dictionaries. make check works as well before
319 so I assume that the patch is correct. I do not know however if things
320 run effectively faster than before now. :)
322 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
323 SimGrid, and the version number is bumped to 2.x
325 2004-10-29 Martin Quinson
326 - Introduction of the remote errors.
327 They are the result of a RMI/RPC on the remote machine.
328 ErrCodes being scalar values, you can't get the host on which those
329 errors did happen. Extending the error mechanism as in Gnome is possible.
330 No idea yet whether it is a good idea.
332 2004-10-28 Martin Quinson
333 - Interface revolution: the Starred Structure Eradication.
334 I used to do typedef struct {} toto_t; and then handle *toto_t.
335 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
336 - ???_t is a valid type (builded with typedef)
337 - s_toto_t is a structure (access to fields with .)
338 - s_toto is a structure needing 'struct' keyword to be used
339 - e_toto_t is an enum
340 - toto_t is an 'object' (struct*)
342 typedef struct s_toto {} s_toto_t, *toto_t;
343 typedef enum {} e_toto_t;
344 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
345 s_toto_t) is private.
347 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
348 so that it can set the variable to NULL. It was so for dicts and sets,
349 it changed for dynars.
351 - Fix a bunch of memleaks in dict_remove
352 - Fix a bug in sg/server_socket opening: it failed all the time.
354 2004-10-07 Martin Quinson
355 - Speed up dynar lookup operation a bit.
357 gras_dynar_get is dead.
359 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
360 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
361 (which gives you the address of the stored data).
363 gras_dynar_get_as is an helpful macro which allows you to retrieve a
364 copy of the data using an affectation to do the job and not a memcpy.
366 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
367 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
369 It does not really speedup the dynar test because they are
370 setting elements all the time (and look them seldom). But the dict does
371 far more lookup than setting.
373 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
375 2004-10-05 Martin Quinson
376 - Allow to (en/dis)able the cycle detection at run time.
378 Whether we should check for cycle or not is now a property of each
379 datatype. When you think there may be some cycle, use datadesc_cycle_set.
380 datadesc_cycle_unset allow to remove this property when previously set.
382 Note that the cycle detection is off by default since it impacts the
383 performance. Watch the data you feed GRAS with ;)
385 This property is hereditary. Any element embedded in a structure having it
386 set have it set for the time of this data exchange.
388 You should set it both on sender and receiver side. If you don't set it on
389 sender side, it will enter an endless loop. If you forget on receiver
390 side, the cycles won't be recreated after communication.
392 - Header reorganization.
393 Kill gras_private.h, each submodule must load the headers it needs.
395 2004-10-04 Martin Quinson
396 - Interface revolution: do not try to survive to malloc failure.
398 Now, gras_malloc and friends call gras_abort() on failure.
399 As a conclusion, malloc_error is not a valid error anymore, and all
400 functions for which it was the only gras_error_t return value are
401 changed. They now return void, or there result directly.
402 This simplify the API a lot.
404 2004-09-29 Martin Quinson
405 - Re-enable raw sockets.
406 Created by gras_socket_{client,server}_ext;
407 Used with gras_raw_{send,recv}
410 It should allow to kill the last bits of gras first version soon.
412 This is not completely satisfactory yet (duplicate code with
413 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
416 - Simplify transport plugin (internal) interface by not passing any
417 argument to _server and _client, but embedding them in the socket
420 2004-09-28 Martin Quinson
421 - Finish the port to AIX.
422 autoconf was my problem (segfault within the malloc replacement
423 function. No idea why)
425 2004-09-16 Martin Quinson
426 - Fix some size_t madness on 64bit architectures.
428 2004-09-08 Martin Quinson
429 - Reduce the number of system headers loaded, overload some more system
430 calls (such as malloc to cast the result of the system one, and work
432 - Fix and reintroduce the config support
434 2004-09-07 Martin Quinson
435 - Source code reorganization to allow Arnaud to surf all over there.
436 - Allow to document the logging categories.
437 - Remove all uppercase from logging categories and useless cleanup in names.
439 2004-08-18 Martin Quinson
440 Version 0.6.2 (protocol not changed; API changed)
441 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
442 gras_error_t), and NULL when not found. Functions expecting a msgtype
443 as argument (msg_wait; msg_send) deal with NULL argument by providing a
444 hopefully usefull message.
445 - Portability to prehistoric sparcs again
447 2004-08-17 Martin Quinson
448 Version 0.6.1 (protocol not changed; ABI not changed)
449 - prealloc some buffers to speed things up
451 2004-08-11 Martin Quinson
452 Version 0.6 (protocol not changed; ABI expended)
453 - The parsing macro can deal with the references, provided that you add
454 the relevant annotations (using GRAS_ANNOTE(size,field_name))
456 2004-08-09 Martin Quinson
457 Version 0.5 (protocol not changed; ABI changed)
458 - Allow to off turn the cycle detection code in data exchange at
459 compilation time. It should be at run time, but I'm short of time (and
460 the config stuff is still broken). That way, we keep dict out of the
461 critical path, which is good because the performance is poor:
462 - search not dichotomial yet
463 - dynar give no way to access their content and memcpy everytime
464 - In composed data description (struct, ref and so on), stop foolness of
465 keeping the subtype's ID, but store the type itself. This keeps sets out
466 of the critical path, which is good since they rely on dynar and
467 dictionnaries. The only loose of that is that we cannot detect the
468 redeclaration of a structure/union with another content (but I'm not sure
469 the code detected well this error before anyway). We still can detect
470 the redefinition discrepancy for the other types.
471 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
472 it breaks the code because of type-punning used all over the place).
473 This breaks on all non-gcc architectures (for now).
475 All those changes (plus the buffer of last time) allow me to gain 2 order
476 of magnitude on cruel tests consisting of 800000 array of integers on two
477 level of a hierarchical structure (200 secondes -> 4 secondes)
480 - the selector of reference must now return the type it points to, not
483 2004-08-06 Martin Quinson
484 Version 0.4 (protocol changed; ABI not changed)
485 - Allow to pass --gras-log argument to processes in simulation mode. Really.
486 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
487 - Implement a buffer transport, and use it by default (it relies on tcp in
488 real life and on sg in simulation).
489 That's a bit hackish since I had a new field to the structure to store
490 its data without interfering with the subtype ones. Inheritance
491 is tricky in C. And that's a kind of reverse inheritance with one class
492 derivating two classes. Or maybe a game with java interfaces. Anyway,
493 that's damn hard in C (at least).
494 Moreover, I got tired while trying to ensure plugin separation and
495 genericity in SG mode. MSG wants me to do weird things, so let's go for
496 cruel hacks (temporarily of course ;).
497 See comment in transport_private.h:71
498 - do not include all the _interface headers in private but in the files
499 which really need them (to cut the compilation time when they are
502 2004-07-26 Martin Quinson
503 Version 0.3 (protocol not changed; ABI changed)
504 - Major overhault of the datadesc interface to simplify it:
505 - shorted the function names:
506 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
507 - add a trivial way to push/pop integers into the cbps without malloc.
508 This allows to make really generic sub_type description, which simply
509 pop their size of the stack.
510 - add a function gras_datadesc_ref_pop_arr() which does what users want
511 most of the time: Declare a dynamic array (which pops its size of the
512 stack) and declare a reference to it. Poor name, but anyway.
513 - kill the post-send callback, add a post-receive one
515 2004-07-23 Martin Quinson
516 Version 0.2 (protocol changed; ABI changed)
517 - add some testing for cpbs in the test cases, and fix some more bugs.
518 This invalidate again the little64 data file, since I cannot regenerate
520 - remove an awfull optimization in the logging stuff, allowing me to:
521 - understand it again
522 - learn gcc how to check that the argument match the provided format
523 - fix all errors revealed by gcc after that
524 - internal keys of dict are not \0 terminated. Deal with it properly in
525 loggings instead of segfaulting when the user want to see the logs :-/
527 2004-07-22 Martin Quinson
528 - Fix some stupid bug preventing cbps (callback postit) from working
530 2004-07-21 Martin Quinson
531 - Some documentation cleanups
532 - remove the useless last argument of msgtype_declare
533 - rename the Virtu functions to fit into the 'os' namespace
534 - move headers src/include -> src/include/gras/ and stop fooling with
535 gras -> . symbolic link
536 - make distcheck is now successful
538 2004-07-19 Martin Quinson
540 - Build shared library also
541 - Install html doc to the right location
542 - stop removing maintainer files in make clean
543 - build tests only on make check
545 2004-07-13 Martin Quinson
547 - No major issue in previous version => change versionning schema
548 - Re-enable little64 convertion test now that Abdou kindly regenerated the
549 corresponding dataset.
551 2004-07-11 Martin Quinson
553 - Get it working with any kind of structure (we can compute the padding
554 bytes remotely for all the architectures I have access to)
555 - Implement the structure parsing macro (still not quite robust/complete)
556 - Improvement to the remote testing toysuite
558 2004-07-10 Martin Quinson
560 - get ride of a bunch of deprecated macros
561 - actually run the test for two-compliment, not only compile it :-/
562 - test whether the structures get packed (and bail out if yes. Damn.
563 Alignment is a serious matter)
564 - test whether the structures get compacted (but respecting the alignment
565 constraints of each types)
566 - test whether the array fields of structures can straddle alignment boundaries
568 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
569 (took me 2 hours to find that bug, looking at the wrong place)
570 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
571 before sending/receiving and are used to compute the offsets of fields
572 - Given that padding size depend even on compiler options, keep track of
573 alignment and aligned_size only for the current architecture. Not a big
574 deal since we send structure fields one after the other (seems
576 - Add the datastructure used for IEEE paper by the PBIO guys to the test
577 program, let it work on linux/gcc/little32. portability todo.
579 2004-07-08 Martin Quinson
580 - import and improve remote compilation support from FAST
581 - make sure make check works on half a dozen of machines out there
583 2004-07-07 Martin Quinson
584 Let's say it's version 0.0.3 ;)
585 - Implement conversions (yuhu!)
586 - Let it work on solaris (beside conversion, of course)
587 - Stupid me, using rand() to generate the conversion datatests in not wise.
589 2004-07-06 Martin Quinson
590 - Let make dist work, since I'm gonna need it to compile on remote hosts
591 - Let Tests/datadesc_usage write the architecture on which the file was
592 generated as first byte.
593 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
594 architecture descriptions.
595 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
596 on those architectures.
597 - Optimization: send/recv array of scalar in one shoot
599 2004-07-05 Martin Quinson
600 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
602 - Plug a whole bunch of memleaks
603 - each process now have to call gras_{init,exit}. One day, their log
604 settings will be separated
605 - Continue the code factorisation between SG, RL and common in Transport.
607 2004-07-04 Martin Quinson
609 - Redistribution between SG and RL.
610 We wanna have to accept in SG, so move accepted related parts of RL in
611 the common part. (more precisely, the dynar of all known sockets is no
612 more a static in transport.c, but part of the process_data)
614 [gras_stub_generator]
615 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
617 2004-07-03 Martin Quinson
618 - Create a new log channel tbx containing dict, set, log, dynar (to shut
619 them all up in one shot)
621 - Fix the ugly case of reference to dynamic array.
622 - New (semi-public) function gras_datadesc_size to allow the messaging
623 layer to malloc the needed space for the buffer.
625 - gras_socket_close now expect the socket to close (and not its address to
626 put NULL in it after it). This is because the socket passed to handlers
627 is one of their argument (=> not writable).
629 - propagate the interface cleanup from last week in datadesc, ie remove a
630 superfluous level of indirection. User pass adress of variable
631 containing data (both when sending and receiving), and not of a variable
632 being a pointer to the data. Let's say that I like it better ;)
633 The price for that is constructs like "int msg=*(int*)payload" in
634 handlers, but it's a fine price, IMHO.
636 - Let it work in RL (yuhu)
638 2004-06-21 Martin Quinson
640 - porting SG plugin and SG select to new standards (works almost).
641 - plug memleaks and fix bugs around.
644 - cleanup the prototype of data recv and force users to specify when they
645 want to handle references to objects. Test case working even for cycles.
646 - plug memleaks. Valgrind is perfectly ok with this.
648 2004-06-12 Martin Quinson
650 - cleanup the separation between plugin and main code in plugin creation
652 2004-06-11 Martin Quinson
654 - Reput hook for raw sockets, needed for BW experiments
655 - kill a few lines of dead code
656 [Data description] Interface cleanup
657 - gras_datadesc_by_name returns the searched type or NULL.
658 That way, no variable is needed to use a type desc once, which makes
660 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
661 two parameters were strings (field name, type name), leading to
663 [Dicos] Interface cleanup
664 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
665 This is consistant with the dynar API.
667 2004-04-21 Martin Quinson
669 - Porting to new standards.
672 There is no bag anymore, no need to take extra provision to mask the
673 pointers behind "ID".
674 Better splitup of functions between files create/exchange/convert.
675 This is still a bit artificial since convert and receive are so
676 interleaved, but anyway.
678 - add a queued message list to procdata (the ones not matching criteria
680 - factorize some more code between SG and RL wrt procdata
682 - use gras_exit in example to track memleaks
683 - get rid of gs_example now that GS is properly integrated into gras
684 - update run_test to integrate the lastest tests (datadesc)
686 - rename WARNINGn macros to WARNn since it prooved error-prone
688 2004-04-19 Martin Quinson
690 - register init/exit functions within gras module mechanism
691 - send/receive function.
692 Convertion is not implemented, but short-cutted if not needed.
693 struct/array elements are sent one by one (instead of block-wise), but
694 nobody really cares (yet). Get a prototype before optimizing.
695 - tests (using a file socket) for DD send/receive on:
696 - base types: int, float
697 - array: fixed size, string (ie ref to dynamic string)
698 - structure: homogeneous, heterogeneous
699 - chained list, graph with cycle
700 Believe it or not, valgrind is not too unhappy with the results. The
701 cycle happily segfaults, but the others are ok. And I'm sick of pointers
705 - Bugfix when using a filename explicitely (instead of '-')
707 2004-04-09 Martin Quinson
709 - factorize more code between RL and SG in socket creation
710 - Complete the implementation and tests of:
712 o file (only in RL, and mainly for debugging)
714 I lost 3 days to design a portable address resolver, and then decided
715 that the prototype mainly have to run on my box.
716 Addressing portability too early may be like optimizing too early :-/
718 - use gras_init in the Tests instead of the crappy parse_log_opt
719 (the latter function is removed)
720 [Conditional execution]
721 - New functions: gras_if_RL/gras_if_SG (basic support for this)
722 [Code reorganisation]
723 - Get rid of libgrasutils.a since it makes more trouble than it solves.
724 Build examples against the RL library, since there is no way to disable
725 its creation for now.