1 SimGrid (2.95) unstable; urgency=low
4 * Implement MSG_get_msgload [MQ]
7 * Bugfix: Allow absolute paths for platform description files [MQ]
10 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
13 * bugfix in automatic type parsing: differentiate the types "char[22]"
14 and "unsigned char[22]". "short" and "long" modifiers were also ignored;
15 other modifier (such as reference level) are still ignored. [MQ]
16 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
20 SimGrid (2.94) unstable; urgency=low
22 The first release candidate for SimGrid 3 !
26 * Update the main page and the FAQ. Adding references to gforge.
29 * Add a gras_os_getpid function.
32 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
33 * Extend the logs so that they also print PID, hostname, date, ... if
35 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
36 and kill the old version which were in testsuite/
37 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
41 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
42 for users. Instead, each programming environment (GRAS, MSG,...) use it
43 when needed to provide such a feature to users.
44 Don't shortcut the mecanism or you will also shortcut the virtualization
45 you need on the simulator.
49 * Cleanups in configury with regard to compile optimization/warning flags.
50 Also add -fno-loop-optimize to any powerpc since it's the optimization
51 killing gcc (< 3.4.0).
52 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
53 needed by MSG examples complications
54 * Borrow configury beautifications from PHP
57 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
58 warning (thanks loris for stumbling into this one).
59 * Bugfix: stop loading private headers (gras_config.h) from the public
63 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
64 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
65 RL would imply a DNS resolver, which is *hard* to do in a portable way
66 (and therefore delayed).
67 * Implement a real timer mecanism and use it in timing macros. This allows
68 to avoid rounding errors and get a 0.000005 sec precision in timing
69 macros. While I was at it, various cleanups:
70 - allow to declare more than one timed section per file (fix a stupid bug)
71 - move some private declaration to the right place
72 - merge conditional execution and timing macros into emulation module
74 - make sure the module cleanups its mess on gras_exit
75 * Documentation improvements:
76 - (new) how to compile applications using GRAS
77 - (new) emulation support (timing macros)
79 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
81 SimGrid (2.93) unstable; urgency=low
83 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
86 - Use Paje properly where used. Still to be sanitized properly.
87 - Portability fix: Add an implementation of the contexts using pthread
91 - Add xbt_procname(): returns the name of the current process.
92 Use it to show the current process's name in all logging.
94 - fix detection of older flex version and the reaction, since we do
95 depend on modern ones (we use lex_destroy)
96 - Better separation of SG and RL in the libs: remove all simulation code
97 from libgras. As a result, this lib is now only 200k when stripped.
98 Some of the xbt modules may also be duplicated (two sets and such) and
99 should be cleaned/killed before SG3.
100 - Insist on using xlC on AIX because of weird problems involving gcc there.
101 - Cleanup the make remote stuff. This is now done by scripts
102 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
103 mainly for our private use, but we're working on changing them to user
106 - Bugfix: flush the socket on close only if there is some *output*.
107 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
109 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
112 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
114 SimGrid (2.92) unstable; urgency=low
116 Alpha 3 on the path to SimGrid 3
120 - New! First try of benchmarking macros.
121 - New! First try so that gras_stub_generator generate deployment and
122 remote compilation helpers.
124 - Bugfix: Initialization fix in msg_test.
128 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
130 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
132 - API Change: gras_os_sleep to take the amount of seconds as a double.
133 Accepting an int was error prone since it was the only location where
134 seconds were coded as such. It leaded to damn rounding errors.
135 - Bugfix: Hard to belive that timers ever worked before this.
137 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
139 SimGrid (2.91) unstable; urgency=low
141 Alpha 2 on the path to SimGrid 3
145 - Bug fix in the lmm_solver.
147 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
148 through the function MSG_paje_output.
149 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
150 - It is possible to bound the rate of a communication in MSG with
151 MSG_task_put_bounded() (was already in the previous version but I had forgotten
152 to write it in the changelog).
153 - Bug fix to let GRAS run on top of MSG until we move it directly on top
158 - Various cleanups to the autotools stuff
159 - Begin to move Gras examples to examples/gras/
160 - Let make distcheck work again (yeah!)
162 - documentation overhauled using doxygen.
163 gtk-doc-tools is dead in SimGrid now.
164 - Automatically extract all existing logging categories, and add the list
165 to the documentation (long standing one, to say the less)
167 - Cleanup the known architecture table. Reorder the entries to group what
168 should be, and use a more consistent naming scheme.
169 (some of the test dataset are still to be regenerated)
170 - New! Allow library to register globals on each process just as userdata
172 This is implemented using a xbt_dict and not a xbt_set, so we loose the
173 lookup time (for now).
174 Use it in msg and trp.
175 This cleans a lot the internals and helps enforcing privacy of the
176 headers between the gras components.
177 - New! Add a timer mechanism, not unlike cron(8) and at(1).
178 - Bugfix: gras_os_time was delirious in RL.
179 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
181 - Reenable GRAS now that it works.
183 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
185 SimGrid (2.90) unstable; urgency=low
187 Alpha 1 on the path to SimGrid 3
189 * It is a long time since the last release of SimGrid. I'm sorry about
190 that but as I had told you, I was rewriting a lot of things. I apologize
191 to those who had been reporting bugs to me and that I had not answered.
192 If your bug is still in the new version, please tell me. Here is a
193 summary of the main changes.
195 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
196 lead by Martin Quinson. As a consequence SimGrid gains a lot in
197 portability, speed, and a lot more but you'll figure it out later.
198 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
199 I wanted to release the new MSG as soon as possible and I have
200 broken GRAS, which is the reason why, for now, only MSG is fully
201 functional. A laconic description of these projects is available
202 in the documentation.
204 * REVOLUTION 2: I have removed SG and I am now using a new simulation
205 kernel optimized for our needs (called SURF but only the developers
206 should use it). Hence, MSG is now roughly 30 times faster and I think
207 that by rewriting a little bit MSG, I could event speed it up a little
208 bit more. Beside the gain in speed, it is also much easier to encode a
209 new platform model with SURF than it was with SG. More to come...
211 * REVOLUTION 3: I have tried to change a little as possible the API of
212 MSG but a few things really had to disappear. The main differences
213 with the previous version are :
214 1) no more m_links_t and the corresponding functions. Platforms are
215 directly read from a XML description and cannot be hard-coded
216 anymore. The same format is used for application deployment
217 description. The new format is described in the documentation.
218 Have a look in tools/platform_generation. There is a tiny script
219 that converts from the old platform format to the new one. Concerning
220 the application deployment format, parsing the old one is tricky.
221 I think most of you should however be able to convert your files. If
222 it is really an issue, I can write a C code that does the conversion.
224 2) the toolbox tbx does not exist anymore. We now have a library
225 with much more data-structures but without the hash-tables (we have
226 dictionaries that are much faster).
228 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
230 *****************************************************************************
231 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
232 * don't feel like converting the oldies. (Mt) *
233 *****************************************************************************
236 Version 2.90: "the long awaited one"
237 - Finished rewriting and debugging MSG. Rewrote the documentation.
238 - disable GRAS for now since it needs to be ported to the newest SG
241 - Finish the port to windows (using mingw32 for cross-compile)
244 - Main loop and datastructures of SURF. A cpu resource object is
245 functional. Surf can thus be used to create cpu's with variable
246 performance on which you can execute some actions.
248 2004-11-15 Martin Quinson
249 - Port to ARM. Simply added the alignment and size descriptions. Should
250 work, but the ARM machines are so slow that I didn't had the opportunity
251 to 'make check' over there yet.
253 2004-11-15 Arnaud Legrand
254 - Trace manager now written. It uses a heap structure and is therefore
255 expected to be efficient. It may however be speeded up (particularly
256 when many events occur at the same date) by using red and black
257 trees. One day maybe...
258 - Max-min linear system solver written. It uses a sparse matrix
259 structure taking advantage of its expected use. Most operations are
260 O(1) and free/calloc are called as few as possible. The computation of
261 the minimum could however be improved by using a red and black tree
264 2004-11-03 Arnaud Legrand
265 - Rename every gras_* function that was in xbt/ to its xbt_
267 - Add a heap and a doubly-linked list to xbt
268 - Added a dichotomy to the dictionaries. make check works as well before
269 so I assume that the patch is correct. I do not know however if things
270 run effectively faster than before now. :)
272 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
273 SimGrid, and the version number is bumped to 2.x
275 2004-10-29 Martin Quinson
276 - Introduction of the remote errors.
277 They are the result of a RMI/RPC on the remote machine.
278 ErrCodes being scalar values, you can't get the host on which those
279 errors did happen. Extending the error mechanism as in Gnome is possible.
280 No idea yet whether it is a good idea.
282 2004-10-28 Martin Quinson
283 - Interface revolution: the Starred Structure Eradication.
284 I used to do typedef struct {} toto_t; and then handle *toto_t.
285 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
286 - ???_t is a valid type (builded with typedef)
287 - s_toto_t is a structure (access to fields with .)
288 - s_toto is a structure needing 'struct' keyword to be used
289 - e_toto_t is an enum
290 - toto_t is an 'object' (struct*)
292 typedef struct s_toto {} s_toto_t, *toto_t;
293 typedef enum {} e_toto_t;
294 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
295 s_toto_t) is private.
297 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
298 so that it can set the variable to NULL. It was so for dicts and sets,
299 it changed for dynars.
301 - Fix a bunch of memleaks in dict_remove
302 - Fix a bug in sg/server_socket opening: it failed all the time.
304 2004-10-07 Martin Quinson
305 - Speed up dynar lookup operation a bit.
307 gras_dynar_get is dead.
309 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
310 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
311 (which gives you the address of the stored data).
313 gras_dynar_get_as is an helpful macro which allows you to retrieve a
314 copy of the data using an affectation to do the job and not a memcpy.
316 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
317 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
319 It does not really speedup the dynar test because they are
320 setting elements all the time (and look them seldom). But the dict does
321 far more lookup than setting.
323 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
325 2004-10-05 Martin Quinson
326 - Allow to (en/dis)able the cycle detection at run time.
328 Whether we should check for cycle or not is now a property of each
329 datatype. When you think there may be some cycle, use datadesc_cycle_set.
330 datadesc_cycle_unset allow to remove this property when previously set.
332 Note that the cycle detection is off by default since it impacts the
333 performance. Watch the data you feed GRAS with ;)
335 This property is hereditary. Any element embedded in a structure having it
336 set have it set for the time of this data exchange.
338 You should set it both on sender and receiver side. If you don't set it on
339 sender side, it will enter an endless loop. If you forget on receiver
340 side, the cycles won't be recreated after communication.
342 - Header reorganization.
343 Kill gras_private.h, each submodule must load the headers it needs.
345 2004-10-04 Martin Quinson
346 - Interface revolution: do not try to survive to malloc failure.
348 Now, gras_malloc and friends call gras_abort() on failure.
349 As a conclusion, malloc_error is not a valid error anymore, and all
350 functions for which it was the only gras_error_t return value are
351 changed. They now return void, or there result directly.
352 This simplify the API a lot.
354 2004-09-29 Martin Quinson
355 - Re-enable raw sockets.
356 Created by gras_socket_{client,server}_ext;
357 Used with gras_raw_{send,recv}
360 It should allow to kill the last bits of gras first version soon.
362 This is not completely satisfactory yet (duplicate code with
363 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
366 - Simplify transport plugin (internal) interface by not passing any
367 argument to _server and _client, but embedding them in the socket
370 2004-09-28 Martin Quinson
371 - Finish the port to AIX.
372 autoconf was my problem (segfault within the malloc replacement
373 function. No idea why)
375 2004-09-16 Martin Quinson
376 - Fix some size_t madness on 64bit architectures.
378 2004-09-08 Martin Quinson
379 - Reduce the number of system headers loaded, overload some more system
380 calls (such as malloc to cast the result of the system one, and work
382 - Fix and reintroduce the config support
384 2004-09-07 Martin Quinson
385 - Source code reorganization to allow Arnaud to surf all over there.
386 - Allow to document the logging categories.
387 - Remove all uppercase from logging categories and useless cleanup in names.
389 2004-08-18 Martin Quinson
390 Version 0.6.2 (protocol not changed; API changed)
391 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
392 gras_error_t), and NULL when not found. Functions expecting a msgtype
393 as argument (msg_wait; msg_send) deal with NULL argument by providing a
394 hopefully usefull message.
395 - Portability to prehistoric sparcs again
397 2004-08-17 Martin Quinson
398 Version 0.6.1 (protocol not changed; ABI not changed)
399 - prealloc some buffers to speed things up
401 2004-08-11 Martin Quinson
402 Version 0.6 (protocol not changed; ABI expended)
403 - The parsing macro can deal with the references, provided that you add
404 the relevant annotations (using GRAS_ANNOTE(size,field_name))
406 2004-08-09 Martin Quinson
407 Version 0.5 (protocol not changed; ABI changed)
408 - Allow to off turn the cycle detection code in data exchange at
409 compilation time. It should be at run time, but I'm short of time (and
410 the config stuff is still broken). That way, we keep dict out of the
411 critical path, which is good because the performance is poor:
412 - search not dichotomial yet
413 - dynar give no way to access their content and memcpy everytime
414 - In composed data description (struct, ref and so on), stop foolness of
415 keeping the subtype's ID, but store the type itself. This keeps sets out
416 of the critical path, which is good since they rely on dynar and
417 dictionnaries. The only loose of that is that we cannot detect the
418 redeclaration of a structure/union with another content (but I'm not sure
419 the code detected well this error before anyway). We still can detect
420 the redefinition discrepancy for the other types.
421 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
422 it breaks the code because of type-punning used all over the place).
423 This breaks on all non-gcc architectures (for now).
425 All those changes (plus the buffer of last time) allow me to gain 2 order
426 of magnitude on cruel tests consisting of 800000 array of integers on two
427 level of a hierarchical structure (200 secondes -> 4 secondes)
430 - the selector of reference must now return the type it points to, not
433 2004-08-06 Martin Quinson
434 Version 0.4 (protocol changed; ABI not changed)
435 - Allow to pass --gras-log argument to processes in simulation mode. Really.
436 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
437 - Implement a buffer transport, and use it by default (it relies on tcp in
438 real life and on sg in simulation).
439 That's a bit hackish since I had a new field to the structure to store
440 its data without interfering with the subtype ones. Inheritance
441 is tricky in C. And that's a kind of reverse inheritance with one class
442 derivating two classes. Or maybe a game with java interfaces. Anyway,
443 that's damn hard in C (at least).
444 Moreover, I got tired while trying to ensure plugin separation and
445 genericity in SG mode. MSG wants me to do weird things, so let's go for
446 cruel hacks (temporarily of course ;).
447 See comment in transport_private.h:71
448 - do not include all the _interface headers in private but in the files
449 which really need them (to cut the compilation time when they are
452 2004-07-26 Martin Quinson
453 Version 0.3 (protocol not changed; ABI changed)
454 - Major overhault of the datadesc interface to simplify it:
455 - shorted the function names:
456 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
457 - add a trivial way to push/pop integers into the cbps without malloc.
458 This allows to make really generic sub_type description, which simply
459 pop their size of the stack.
460 - add a function gras_datadesc_ref_pop_arr() which does what users want
461 most of the time: Declare a dynamic array (which pops its size of the
462 stack) and declare a reference to it. Poor name, but anyway.
463 - kill the post-send callback, add a post-receive one
465 2004-07-23 Martin Quinson
466 Version 0.2 (protocol changed; ABI changed)
467 - add some testing for cpbs in the test cases, and fix some more bugs.
468 This invalidate again the little64 data file, since I cannot regenerate
470 - remove an awfull optimization in the logging stuff, allowing me to:
471 - understand it again
472 - learn gcc how to check that the argument match the provided format
473 - fix all errors revealed by gcc after that
474 - internal keys of dict are not \0 terminated. Deal with it properly in
475 loggings instead of segfaulting when the user want to see the logs :-/
477 2004-07-22 Martin Quinson
478 - Fix some stupid bug preventing cbps (callback postit) from working
480 2004-07-21 Martin Quinson
481 - Some documentation cleanups
482 - remove the useless last argument of msgtype_declare
483 - rename the Virtu functions to fit into the 'os' namespace
484 - move headers src/include -> src/include/gras/ and stop fooling with
485 gras -> . symbolic link
486 - make distcheck is now successful
488 2004-07-19 Martin Quinson
490 - Build shared library also
491 - Install html doc to the right location
492 - stop removing maintainer files in make clean
493 - build tests only on make check
495 2004-07-13 Martin Quinson
497 - No major issue in previous version => change versionning schema
498 - Re-enable little64 convertion test now that Abdou kindly regenerated the
499 corresponding dataset.
501 2004-07-11 Martin Quinson
503 - Get it working with any kind of structure (we can compute the padding
504 bytes remotely for all the architectures I have access to)
505 - Implement the structure parsing macro (still not quite robust/complete)
506 - Improvement to the remote testing toysuite
508 2004-07-10 Martin Quinson
510 - get ride of a bunch of deprecated macros
511 - actually run the test for two-compliment, not only compile it :-/
512 - test whether the structures get packed (and bail out if yes. Damn.
513 Alignment is a serious matter)
514 - test whether the structures get compacted (but respecting the alignment
515 constraints of each types)
516 - test whether the array fields of structures can straddle alignment boundaries
518 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
519 (took me 2 hours to find that bug, looking at the wrong place)
520 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
521 before sending/receiving and are used to compute the offsets of fields
522 - Given that padding size depend even on compiler options, keep track of
523 alignment and aligned_size only for the current architecture. Not a big
524 deal since we send structure fields one after the other (seems
526 - Add the datastructure used for IEEE paper by the PBIO guys to the test
527 program, let it work on linux/gcc/little32. portability todo.
529 2004-07-08 Martin Quinson
530 - import and improve remote compilation support from FAST
531 - make sure make check works on half a dozen of machines out there
533 2004-07-07 Martin Quinson
534 Let's say it's version 0.0.3 ;)
535 - Implement conversions (yuhu!)
536 - Let it work on solaris (beside conversion, of course)
537 - Stupid me, using rand() to generate the conversion datatests in not wise.
539 2004-07-06 Martin Quinson
540 - Let make dist work, since I'm gonna need it to compile on remote hosts
541 - Let Tests/datadesc_usage write the architecture on which the file was
542 generated as first byte.
543 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
544 architecture descriptions.
545 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
546 on those architectures.
547 - Optimization: send/recv array of scalar in one shoot
549 2004-07-05 Martin Quinson
550 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
552 - Plug a whole bunch of memleaks
553 - each process now have to call gras_{init,exit}. One day, their log
554 settings will be separated
555 - Continue the code factorisation between SG, RL and common in Transport.
557 2004-07-04 Martin Quinson
559 - Redistribution between SG and RL.
560 We wanna have to accept in SG, so move accepted related parts of RL in
561 the common part. (more precisely, the dynar of all known sockets is no
562 more a static in transport.c, but part of the process_data)
564 [gras_stub_generator]
565 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
567 2004-07-03 Martin Quinson
568 - Create a new log channel tbx containing dict, set, log, dynar (to shut
569 them all up in one shot)
571 - Fix the ugly case of reference to dynamic array.
572 - New (semi-public) function gras_datadesc_size to allow the messaging
573 layer to malloc the needed space for the buffer.
575 - gras_socket_close now expect the socket to close (and not its address to
576 put NULL in it after it). This is because the socket passed to handlers
577 is one of their argument (=> not writable).
579 - propagate the interface cleanup from last week in datadesc, ie remove a
580 superfluous level of indirection. User pass adress of variable
581 containing data (both when sending and receiving), and not of a variable
582 being a pointer to the data. Let's say that I like it better ;)
583 The price for that is constructs like "int msg=*(int*)payload" in
584 handlers, but it's a fine price, IMHO.
586 - Let it work in RL (yuhu)
588 2004-06-21 Martin Quinson
590 - porting SG plugin and SG select to new standards (works almost).
591 - plug memleaks and fix bugs around.
594 - cleanup the prototype of data recv and force users to specify when they
595 want to handle references to objects. Test case working even for cycles.
596 - plug memleaks. Valgrind is perfectly ok with this.
598 2004-06-12 Martin Quinson
600 - cleanup the separation between plugin and main code in plugin creation
602 2004-06-11 Martin Quinson
604 - Reput hook for raw sockets, needed for BW experiments
605 - kill a few lines of dead code
606 [Data description] Interface cleanup
607 - gras_datadesc_by_name returns the searched type or NULL.
608 That way, no variable is needed to use a type desc once, which makes
610 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
611 two parameters were strings (field name, type name), leading to
613 [Dicos] Interface cleanup
614 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
615 This is consistant with the dynar API.
617 2004-04-21 Martin Quinson
619 - Porting to new standards.
622 There is no bag anymore, no need to take extra provision to mask the
623 pointers behind "ID".
624 Better splitup of functions between files create/exchange/convert.
625 This is still a bit artificial since convert and receive are so
626 interleaved, but anyway.
628 - add a queued message list to procdata (the ones not matching criteria
630 - factorize some more code between SG and RL wrt procdata
632 - use gras_exit in example to track memleaks
633 - get rid of gs_example now that GS is properly integrated into gras
634 - update run_test to integrate the lastest tests (datadesc)
636 - rename WARNINGn macros to WARNn since it prooved error-prone
638 2004-04-19 Martin Quinson
640 - register init/exit functions within gras module mechanism
641 - send/receive function.
642 Convertion is not implemented, but short-cutted if not needed.
643 struct/array elements are sent one by one (instead of block-wise), but
644 nobody really cares (yet). Get a prototype before optimizing.
645 - tests (using a file socket) for DD send/receive on:
646 - base types: int, float
647 - array: fixed size, string (ie ref to dynamic string)
648 - structure: homogeneous, heterogeneous
649 - chained list, graph with cycle
650 Believe it or not, valgrind is not too unhappy with the results. The
651 cycle happily segfaults, but the others are ok. And I'm sick of pointers
655 - Bugfix when using a filename explicitely (instead of '-')
657 2004-04-09 Martin Quinson
659 - factorize more code between RL and SG in socket creation
660 - Complete the implementation and tests of:
662 o file (only in RL, and mainly for debugging)
664 I lost 3 days to design a portable address resolver, and then decided
665 that the prototype mainly have to run on my box.
666 Addressing portability too early may be like optimizing too early :-/
668 - use gras_init in the Tests instead of the crappy parse_log_opt
669 (the latter function is removed)
670 [Conditional execution]
671 - New functions: gras_if_RL/gras_if_SG (basic support for this)
672 [Code reorganisation]
673 - Get rid of libgrasutils.a since it makes more trouble than it solves.
674 Build examples against the RL library, since there is no way to disable
675 its creation for now.