1 SimGrid (2.95) unstable; urgency=low
4 * Steal some nice code to GNU pth to fix context detection and usage [AL]
5 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
6 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
9 * Implement MSG_get_msgload [MQ, AL]
12 * Bugfix: Allow absolute paths for platform description files [MQ]
13 * Bugfix: do free the variables after use. Leads to drastic performance
17 * Implement measurement sockets. You can now get the bandwidth between two
18 hosts thanks to AMOK (see below). [MQ]
19 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
20 dynar over the network (yeah) [MQ]
21 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
22 * simple/static token-ring example. [Alexandre Colucci and MQ]
25 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
26 type parsing. "short" and "long" modifiers were also ignored; other
27 modifier (such as reference level) are still ignored. [MQ]
28 * Embeed the buffer size within the buffer itself on SG. [MQ]
29 That way, send() are atomic and cannot get intermixed anymore (at least
30 the ones which are less than 100k; bigger messages still have the issue)
31 * Array size pushed by the field, not by the field type (or each
32 and every long int will push stuff to the cbps) [MQ]
33 * use select() to sleep since it allows to portably sleep less than one
37 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
40 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
41 * type category 'ignored' killed as it was never used and were difficult
43 * whether a type can cycle or not is now a flag, leaving room for more
44 flags (as "ignored", if we feel the need one day ;)
45 * Rename raw sockets to measurement sockets since "raw" has another
46 meaning in networking community.
49 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
50 and offers features not belonging to GRAS but that most applications
51 need. One day, it may be a set of plugins loadable at runtime.
52 * New module: bandwidth
53 bandwidth measurement between arbitrary nodes running this module. [MQ]
57 SimGrid (2.94) unstable; urgency=low
59 The first beta release of SimGrid 3 !
63 * Update the main page and the FAQ. Adding references to gforge.
66 * Add a gras_os_getpid function.
69 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
70 * Extend the logs so that they also print PID, hostname, date, ... if
72 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
73 and kill the old version which were in testsuite/
74 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
78 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
79 for users. Instead, each programming environment (GRAS, MSG,...) use it
80 when needed to provide such a feature to users.
81 Don't shortcut the mecanism or you will also shortcut the virtualization
82 you need on the simulator.
86 * Cleanups in configury with regard to compile optimization/warning flags.
87 Also add -fno-loop-optimize to any powerpc since it's the optimization
88 killing gcc (< 3.4.0).
89 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
90 needed by MSG examples complications
91 * Borrow configury beautifications from PHP
94 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
95 warning (thanks loris for stumbling into this one).
96 * Bugfix: stop loading private headers (gras_config.h) from the public
100 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
101 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
102 RL would imply a DNS resolver, which is *hard* to do in a portable way
103 (and therefore delayed).
104 * Implement a real timer mecanism and use it in timing macros. This allows
105 to avoid rounding errors and get a 0.000005 sec precision in timing
106 macros. While I was at it, various cleanups:
107 - allow to declare more than one timed section per file (fix a stupid bug)
108 - move some private declaration to the right place
109 - merge conditional execution and timing macros into emulation module
110 - document the module
111 - make sure the module cleanups its mess on gras_exit
112 * Documentation improvements:
113 - (new) how to compile applications using GRAS
114 - (new) emulation support (timing macros)
116 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
118 SimGrid (2.93) unstable; urgency=low
120 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
123 - Use Paje properly where used. Still to be sanitized properly.
124 - Portability fix: Add an implementation of the contexts using pthread
128 - Add xbt_procname(): returns the name of the current process.
129 Use it to show the current process's name in all logging.
131 - fix detection of older flex version and the reaction, since we do
132 depend on modern ones (we use lex_destroy)
133 - Better separation of SG and RL in the libs: remove all simulation code
134 from libgras. As a result, this lib is now only 200k when stripped.
135 Some of the xbt modules may also be duplicated (two sets and such) and
136 should be cleaned/killed before SG3.
137 - Insist on using xlC on AIX because of weird problems involving gcc there.
138 - Cleanup the make remote stuff. This is now done by scripts
139 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
140 mainly for our private use, but we're working on changing them to user
143 - Bugfix: flush the socket on close only if there is some *output*.
144 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
146 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
149 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
151 SimGrid (2.92) unstable; urgency=low
153 Alpha 3 on the path to SimGrid 3
157 - New! First try of benchmarking macros.
158 - New! First try so that gras_stub_generator generate deployment and
159 remote compilation helpers.
161 - Bugfix: Initialization fix in msg_test.
165 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
167 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
169 - API Change: gras_os_sleep to take the amount of seconds as a double.
170 Accepting an int was error prone since it was the only location where
171 seconds were coded as such. It leaded to damn rounding errors.
172 - Bugfix: Hard to belive that timers ever worked before this.
174 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
176 SimGrid (2.91) unstable; urgency=low
178 Alpha 2 on the path to SimGrid 3
182 - Bug fix in the lmm_solver.
184 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
185 through the function MSG_paje_output.
186 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
187 - It is possible to bound the rate of a communication in MSG with
188 MSG_task_put_bounded() (was already in the previous version but I had forgotten
189 to write it in the changelog).
190 - Bug fix to let GRAS run on top of MSG until we move it directly on top
195 - Various cleanups to the autotools stuff
196 - Begin to move Gras examples to examples/gras/
197 - Let make distcheck work again (yeah!)
199 - documentation overhauled using doxygen.
200 gtk-doc-tools is dead in SimGrid now.
201 - Automatically extract all existing logging categories, and add the list
202 to the documentation (long standing one, to say the less)
204 - Cleanup the known architecture table. Reorder the entries to group what
205 should be, and use a more consistent naming scheme.
206 (some of the test dataset are still to be regenerated)
207 - New! Allow library to register globals on each process just as userdata
209 This is implemented using a xbt_dict and not a xbt_set, so we loose the
210 lookup time (for now).
211 Use it in msg and trp.
212 This cleans a lot the internals and helps enforcing privacy of the
213 headers between the gras components.
214 - New! Add a timer mechanism, not unlike cron(8) and at(1).
215 - Bugfix: gras_os_time was delirious in RL.
216 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
218 - Reenable GRAS now that it works.
220 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
222 SimGrid (2.90) unstable; urgency=low
224 Alpha 1 on the path to SimGrid 3
226 * It is a long time since the last release of SimGrid. I'm sorry about
227 that but as I had told you, I was rewriting a lot of things. I apologize
228 to those who had been reporting bugs to me and that I had not answered.
229 If your bug is still in the new version, please tell me. Here is a
230 summary of the main changes.
232 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
233 lead by Martin Quinson. As a consequence SimGrid gains a lot in
234 portability, speed, and a lot more but you'll figure it out later.
235 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
236 I wanted to release the new MSG as soon as possible and I have
237 broken GRAS, which is the reason why, for now, only MSG is fully
238 functional. A laconic description of these projects is available
239 in the documentation.
241 * REVOLUTION 2: I have removed SG and I am now using a new simulation
242 kernel optimized for our needs (called SURF but only the developers
243 should use it). Hence, MSG is now roughly 30 times faster and I think
244 that by rewriting a little bit MSG, I could event speed it up a little
245 bit more. Beside the gain in speed, it is also much easier to encode a
246 new platform model with SURF than it was with SG. More to come...
248 * REVOLUTION 3: I have tried to change a little as possible the API of
249 MSG but a few things really had to disappear. The main differences
250 with the previous version are :
251 1) no more m_links_t and the corresponding functions. Platforms are
252 directly read from a XML description and cannot be hard-coded
253 anymore. The same format is used for application deployment
254 description. The new format is described in the documentation.
255 Have a look in tools/platform_generation. There is a tiny script
256 that converts from the old platform format to the new one. Concerning
257 the application deployment format, parsing the old one is tricky.
258 I think most of you should however be able to convert your files. If
259 it is really an issue, I can write a C code that does the conversion.
261 2) the toolbox tbx does not exist anymore. We now have a library
262 with much more data-structures but without the hash-tables (we have
263 dictionaries that are much faster).
265 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
267 *****************************************************************************
268 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
269 * don't feel like converting the oldies. (Mt) *
270 *****************************************************************************
273 Version 2.90: "the long awaited one"
274 - Finished rewriting and debugging MSG. Rewrote the documentation.
275 - disable GRAS for now since it needs to be ported to the newest SG
278 - Finish the port to windows (using mingw32 for cross-compile)
281 - Main loop and datastructures of SURF. A cpu resource object is
282 functional. Surf can thus be used to create cpu's with variable
283 performance on which you can execute some actions.
285 2004-11-15 Martin Quinson
286 - Port to ARM. Simply added the alignment and size descriptions. Should
287 work, but the ARM machines are so slow that I didn't had the opportunity
288 to 'make check' over there yet.
290 2004-11-15 Arnaud Legrand
291 - Trace manager now written. It uses a heap structure and is therefore
292 expected to be efficient. It may however be speeded up (particularly
293 when many events occur at the same date) by using red and black
294 trees. One day maybe...
295 - Max-min linear system solver written. It uses a sparse matrix
296 structure taking advantage of its expected use. Most operations are
297 O(1) and free/calloc are called as few as possible. The computation of
298 the minimum could however be improved by using a red and black tree
301 2004-11-03 Arnaud Legrand
302 - Rename every gras_* function that was in xbt/ to its xbt_
304 - Add a heap and a doubly-linked list to xbt
305 - Added a dichotomy to the dictionaries. make check works as well before
306 so I assume that the patch is correct. I do not know however if things
307 run effectively faster than before now. :)
309 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
310 SimGrid, and the version number is bumped to 2.x
312 2004-10-29 Martin Quinson
313 - Introduction of the remote errors.
314 They are the result of a RMI/RPC on the remote machine.
315 ErrCodes being scalar values, you can't get the host on which those
316 errors did happen. Extending the error mechanism as in Gnome is possible.
317 No idea yet whether it is a good idea.
319 2004-10-28 Martin Quinson
320 - Interface revolution: the Starred Structure Eradication.
321 I used to do typedef struct {} toto_t; and then handle *toto_t.
322 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
323 - ???_t is a valid type (builded with typedef)
324 - s_toto_t is a structure (access to fields with .)
325 - s_toto is a structure needing 'struct' keyword to be used
326 - e_toto_t is an enum
327 - toto_t is an 'object' (struct*)
329 typedef struct s_toto {} s_toto_t, *toto_t;
330 typedef enum {} e_toto_t;
331 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
332 s_toto_t) is private.
334 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
335 so that it can set the variable to NULL. It was so for dicts and sets,
336 it changed for dynars.
338 - Fix a bunch of memleaks in dict_remove
339 - Fix a bug in sg/server_socket opening: it failed all the time.
341 2004-10-07 Martin Quinson
342 - Speed up dynar lookup operation a bit.
344 gras_dynar_get is dead.
346 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
347 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
348 (which gives you the address of the stored data).
350 gras_dynar_get_as is an helpful macro which allows you to retrieve a
351 copy of the data using an affectation to do the job and not a memcpy.
353 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
354 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
356 It does not really speedup the dynar test because they are
357 setting elements all the time (and look them seldom). But the dict does
358 far more lookup than setting.
360 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
362 2004-10-05 Martin Quinson
363 - Allow to (en/dis)able the cycle detection at run time.
365 Whether we should check for cycle or not is now a property of each
366 datatype. When you think there may be some cycle, use datadesc_cycle_set.
367 datadesc_cycle_unset allow to remove this property when previously set.
369 Note that the cycle detection is off by default since it impacts the
370 performance. Watch the data you feed GRAS with ;)
372 This property is hereditary. Any element embedded in a structure having it
373 set have it set for the time of this data exchange.
375 You should set it both on sender and receiver side. If you don't set it on
376 sender side, it will enter an endless loop. If you forget on receiver
377 side, the cycles won't be recreated after communication.
379 - Header reorganization.
380 Kill gras_private.h, each submodule must load the headers it needs.
382 2004-10-04 Martin Quinson
383 - Interface revolution: do not try to survive to malloc failure.
385 Now, gras_malloc and friends call gras_abort() on failure.
386 As a conclusion, malloc_error is not a valid error anymore, and all
387 functions for which it was the only gras_error_t return value are
388 changed. They now return void, or there result directly.
389 This simplify the API a lot.
391 2004-09-29 Martin Quinson
392 - Re-enable raw sockets.
393 Created by gras_socket_{client,server}_ext;
394 Used with gras_raw_{send,recv}
397 It should allow to kill the last bits of gras first version soon.
399 This is not completely satisfactory yet (duplicate code with
400 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
403 - Simplify transport plugin (internal) interface by not passing any
404 argument to _server and _client, but embedding them in the socket
407 2004-09-28 Martin Quinson
408 - Finish the port to AIX.
409 autoconf was my problem (segfault within the malloc replacement
410 function. No idea why)
412 2004-09-16 Martin Quinson
413 - Fix some size_t madness on 64bit architectures.
415 2004-09-08 Martin Quinson
416 - Reduce the number of system headers loaded, overload some more system
417 calls (such as malloc to cast the result of the system one, and work
419 - Fix and reintroduce the config support
421 2004-09-07 Martin Quinson
422 - Source code reorganization to allow Arnaud to surf all over there.
423 - Allow to document the logging categories.
424 - Remove all uppercase from logging categories and useless cleanup in names.
426 2004-08-18 Martin Quinson
427 Version 0.6.2 (protocol not changed; API changed)
428 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
429 gras_error_t), and NULL when not found. Functions expecting a msgtype
430 as argument (msg_wait; msg_send) deal with NULL argument by providing a
431 hopefully usefull message.
432 - Portability to prehistoric sparcs again
434 2004-08-17 Martin Quinson
435 Version 0.6.1 (protocol not changed; ABI not changed)
436 - prealloc some buffers to speed things up
438 2004-08-11 Martin Quinson
439 Version 0.6 (protocol not changed; ABI expended)
440 - The parsing macro can deal with the references, provided that you add
441 the relevant annotations (using GRAS_ANNOTE(size,field_name))
443 2004-08-09 Martin Quinson
444 Version 0.5 (protocol not changed; ABI changed)
445 - Allow to off turn the cycle detection code in data exchange at
446 compilation time. It should be at run time, but I'm short of time (and
447 the config stuff is still broken). That way, we keep dict out of the
448 critical path, which is good because the performance is poor:
449 - search not dichotomial yet
450 - dynar give no way to access their content and memcpy everytime
451 - In composed data description (struct, ref and so on), stop foolness of
452 keeping the subtype's ID, but store the type itself. This keeps sets out
453 of the critical path, which is good since they rely on dynar and
454 dictionnaries. The only loose of that is that we cannot detect the
455 redeclaration of a structure/union with another content (but I'm not sure
456 the code detected well this error before anyway). We still can detect
457 the redefinition discrepancy for the other types.
458 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
459 it breaks the code because of type-punning used all over the place).
460 This breaks on all non-gcc architectures (for now).
462 All those changes (plus the buffer of last time) allow me to gain 2 order
463 of magnitude on cruel tests consisting of 800000 array of integers on two
464 level of a hierarchical structure (200 secondes -> 4 secondes)
467 - the selector of reference must now return the type it points to, not
470 2004-08-06 Martin Quinson
471 Version 0.4 (protocol changed; ABI not changed)
472 - Allow to pass --gras-log argument to processes in simulation mode. Really.
473 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
474 - Implement a buffer transport, and use it by default (it relies on tcp in
475 real life and on sg in simulation).
476 That's a bit hackish since I had a new field to the structure to store
477 its data without interfering with the subtype ones. Inheritance
478 is tricky in C. And that's a kind of reverse inheritance with one class
479 derivating two classes. Or maybe a game with java interfaces. Anyway,
480 that's damn hard in C (at least).
481 Moreover, I got tired while trying to ensure plugin separation and
482 genericity in SG mode. MSG wants me to do weird things, so let's go for
483 cruel hacks (temporarily of course ;).
484 See comment in transport_private.h:71
485 - do not include all the _interface headers in private but in the files
486 which really need them (to cut the compilation time when they are
489 2004-07-26 Martin Quinson
490 Version 0.3 (protocol not changed; ABI changed)
491 - Major overhault of the datadesc interface to simplify it:
492 - shorted the function names:
493 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
494 - add a trivial way to push/pop integers into the cbps without malloc.
495 This allows to make really generic sub_type description, which simply
496 pop their size of the stack.
497 - add a function gras_datadesc_ref_pop_arr() which does what users want
498 most of the time: Declare a dynamic array (which pops its size of the
499 stack) and declare a reference to it. Poor name, but anyway.
500 - kill the post-send callback, add a post-receive one
502 2004-07-23 Martin Quinson
503 Version 0.2 (protocol changed; ABI changed)
504 - add some testing for cpbs in the test cases, and fix some more bugs.
505 This invalidate again the little64 data file, since I cannot regenerate
507 - remove an awfull optimization in the logging stuff, allowing me to:
508 - understand it again
509 - learn gcc how to check that the argument match the provided format
510 - fix all errors revealed by gcc after that
511 - internal keys of dict are not \0 terminated. Deal with it properly in
512 loggings instead of segfaulting when the user want to see the logs :-/
514 2004-07-22 Martin Quinson
515 - Fix some stupid bug preventing cbps (callback postit) from working
517 2004-07-21 Martin Quinson
518 - Some documentation cleanups
519 - remove the useless last argument of msgtype_declare
520 - rename the Virtu functions to fit into the 'os' namespace
521 - move headers src/include -> src/include/gras/ and stop fooling with
522 gras -> . symbolic link
523 - make distcheck is now successful
525 2004-07-19 Martin Quinson
527 - Build shared library also
528 - Install html doc to the right location
529 - stop removing maintainer files in make clean
530 - build tests only on make check
532 2004-07-13 Martin Quinson
534 - No major issue in previous version => change versionning schema
535 - Re-enable little64 convertion test now that Abdou kindly regenerated the
536 corresponding dataset.
538 2004-07-11 Martin Quinson
540 - Get it working with any kind of structure (we can compute the padding
541 bytes remotely for all the architectures I have access to)
542 - Implement the structure parsing macro (still not quite robust/complete)
543 - Improvement to the remote testing toysuite
545 2004-07-10 Martin Quinson
547 - get ride of a bunch of deprecated macros
548 - actually run the test for two-compliment, not only compile it :-/
549 - test whether the structures get packed (and bail out if yes. Damn.
550 Alignment is a serious matter)
551 - test whether the structures get compacted (but respecting the alignment
552 constraints of each types)
553 - test whether the array fields of structures can straddle alignment boundaries
555 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
556 (took me 2 hours to find that bug, looking at the wrong place)
557 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
558 before sending/receiving and are used to compute the offsets of fields
559 - Given that padding size depend even on compiler options, keep track of
560 alignment and aligned_size only for the current architecture. Not a big
561 deal since we send structure fields one after the other (seems
563 - Add the datastructure used for IEEE paper by the PBIO guys to the test
564 program, let it work on linux/gcc/little32. portability todo.
566 2004-07-08 Martin Quinson
567 - import and improve remote compilation support from FAST
568 - make sure make check works on half a dozen of machines out there
570 2004-07-07 Martin Quinson
571 Let's say it's version 0.0.3 ;)
572 - Implement conversions (yuhu!)
573 - Let it work on solaris (beside conversion, of course)
574 - Stupid me, using rand() to generate the conversion datatests in not wise.
576 2004-07-06 Martin Quinson
577 - Let make dist work, since I'm gonna need it to compile on remote hosts
578 - Let Tests/datadesc_usage write the architecture on which the file was
579 generated as first byte.
580 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
581 architecture descriptions.
582 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
583 on those architectures.
584 - Optimization: send/recv array of scalar in one shoot
586 2004-07-05 Martin Quinson
587 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
589 - Plug a whole bunch of memleaks
590 - each process now have to call gras_{init,exit}. One day, their log
591 settings will be separated
592 - Continue the code factorisation between SG, RL and common in Transport.
594 2004-07-04 Martin Quinson
596 - Redistribution between SG and RL.
597 We wanna have to accept in SG, so move accepted related parts of RL in
598 the common part. (more precisely, the dynar of all known sockets is no
599 more a static in transport.c, but part of the process_data)
601 [gras_stub_generator]
602 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
604 2004-07-03 Martin Quinson
605 - Create a new log channel tbx containing dict, set, log, dynar (to shut
606 them all up in one shot)
608 - Fix the ugly case of reference to dynamic array.
609 - New (semi-public) function gras_datadesc_size to allow the messaging
610 layer to malloc the needed space for the buffer.
612 - gras_socket_close now expect the socket to close (and not its address to
613 put NULL in it after it). This is because the socket passed to handlers
614 is one of their argument (=> not writable).
616 - propagate the interface cleanup from last week in datadesc, ie remove a
617 superfluous level of indirection. User pass adress of variable
618 containing data (both when sending and receiving), and not of a variable
619 being a pointer to the data. Let's say that I like it better ;)
620 The price for that is constructs like "int msg=*(int*)payload" in
621 handlers, but it's a fine price, IMHO.
623 - Let it work in RL (yuhu)
625 2004-06-21 Martin Quinson
627 - porting SG plugin and SG select to new standards (works almost).
628 - plug memleaks and fix bugs around.
631 - cleanup the prototype of data recv and force users to specify when they
632 want to handle references to objects. Test case working even for cycles.
633 - plug memleaks. Valgrind is perfectly ok with this.
635 2004-06-12 Martin Quinson
637 - cleanup the separation between plugin and main code in plugin creation
639 2004-06-11 Martin Quinson
641 - Reput hook for raw sockets, needed for BW experiments
642 - kill a few lines of dead code
643 [Data description] Interface cleanup
644 - gras_datadesc_by_name returns the searched type or NULL.
645 That way, no variable is needed to use a type desc once, which makes
647 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
648 two parameters were strings (field name, type name), leading to
650 [Dicos] Interface cleanup
651 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
652 This is consistant with the dynar API.
654 2004-04-21 Martin Quinson
656 - Porting to new standards.
659 There is no bag anymore, no need to take extra provision to mask the
660 pointers behind "ID".
661 Better splitup of functions between files create/exchange/convert.
662 This is still a bit artificial since convert and receive are so
663 interleaved, but anyway.
665 - add a queued message list to procdata (the ones not matching criteria
667 - factorize some more code between SG and RL wrt procdata
669 - use gras_exit in example to track memleaks
670 - get rid of gs_example now that GS is properly integrated into gras
671 - update run_test to integrate the lastest tests (datadesc)
673 - rename WARNINGn macros to WARNn since it prooved error-prone
675 2004-04-19 Martin Quinson
677 - register init/exit functions within gras module mechanism
678 - send/receive function.
679 Convertion is not implemented, but short-cutted if not needed.
680 struct/array elements are sent one by one (instead of block-wise), but
681 nobody really cares (yet). Get a prototype before optimizing.
682 - tests (using a file socket) for DD send/receive on:
683 - base types: int, float
684 - array: fixed size, string (ie ref to dynamic string)
685 - structure: homogeneous, heterogeneous
686 - chained list, graph with cycle
687 Believe it or not, valgrind is not too unhappy with the results. The
688 cycle happily segfaults, but the others are ok. And I'm sick of pointers
692 - Bugfix when using a filename explicitely (instead of '-')
694 2004-04-09 Martin Quinson
696 - factorize more code between RL and SG in socket creation
697 - Complete the implementation and tests of:
699 o file (only in RL, and mainly for debugging)
701 I lost 3 days to design a portable address resolver, and then decided
702 that the prototype mainly have to run on my box.
703 Addressing portability too early may be like optimizing too early :-/
705 - use gras_init in the Tests instead of the crappy parse_log_opt
706 (the latter function is removed)
707 [Conditional execution]
708 - New functions: gras_if_RL/gras_if_SG (basic support for this)
709 [Code reorganisation]
710 - Get rid of libgrasutils.a since it makes more trouble than it solves.
711 Build examples against the RL library, since there is no way to disable
712 its creation for now.