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]
19 SimGrid (2.94) unstable; urgency=low
21 The first release candidate for SimGrid 3 !
25 * Update the main page and the FAQ. Adding references to gforge.
28 * Add a gras_os_getpid function.
31 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
32 * Extend the logs so that they also print PID, hostname, date, ... if
34 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
35 and kill the old version which were in testsuite/
36 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
40 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
41 for users. Instead, each programming environment (GRAS, MSG,...) use it
42 when needed to provide such a feature to users.
43 Don't shortcut the mecanism or you will also shortcut the virtualization
44 you need on the simulator.
48 * Cleanups in configury with regard to compile optimization/warning flags.
49 Also add -fno-loop-optimize to any powerpc since it's the optimization
50 killing gcc (< 3.4.0).
51 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
52 needed by MSG examples complications
53 * Borrow configury beautifications from PHP
56 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
57 warning (thanks loris for stumbling into this one).
58 * Bugfix: stop loading private headers (gras_config.h) from the public
62 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
63 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
64 RL would imply a DNS resolver, which is *hard* to do in a portable way
65 (and therefore delayed).
66 * Implement a real timer mecanism and use it in timing macros. This allows
67 to avoid rounding errors and get a 0.000005 sec precision in timing
68 macros. While I was at it, various cleanups:
69 - allow to declare more than one timed section per file (fix a stupid bug)
70 - move some private declaration to the right place
71 - merge conditional execution and timing macros into emulation module
73 - make sure the module cleanups its mess on gras_exit
74 * Documentation improvements:
75 - (new) how to compile applications using GRAS
76 - (new) emulation support (timing macros)
78 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
80 SimGrid (2.93) unstable; urgency=low
82 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
85 - Use Paje properly where used. Still to be sanitized properly.
86 - Portability fix: Add an implementation of the contexts using pthread
90 - Add xbt_procname(): returns the name of the current process.
91 Use it to show the current process's name in all logging.
93 - fix detection of older flex version and the reaction, since we do
94 depend on modern ones (we use lex_destroy)
95 - Better separation of SG and RL in the libs: remove all simulation code
96 from libgras. As a result, this lib is now only 200k when stripped.
97 Some of the xbt modules may also be duplicated (two sets and such) and
98 should be cleaned/killed before SG3.
99 - Insist on using xlC on AIX because of weird problems involving gcc there.
100 - Cleanup the make remote stuff. This is now done by scripts
101 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
102 mainly for our private use, but we're working on changing them to user
105 - Bugfix: flush the socket on close only if there is some *output*.
106 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
108 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
111 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
113 SimGrid (2.92) unstable; urgency=low
115 Alpha 3 on the path to SimGrid 3
119 - New! First try of benchmarking macros.
120 - New! First try so that gras_stub_generator generate deployment and
121 remote compilation helpers.
123 - Bugfix: Initialization fix in msg_test.
127 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
129 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
131 - API Change: gras_os_sleep to take the amount of seconds as a double.
132 Accepting an int was error prone since it was the only location where
133 seconds were coded as such. It leaded to damn rounding errors.
134 - Bugfix: Hard to belive that timers ever worked before this.
136 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
138 SimGrid (2.91) unstable; urgency=low
140 Alpha 2 on the path to SimGrid 3
144 - Bug fix in the lmm_solver.
146 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
147 through the function MSG_paje_output.
148 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
149 - It is possible to bound the rate of a communication in MSG with
150 MSG_task_put_bounded() (was already in the previous version but I had forgotten
151 to write it in the changelog).
152 - Bug fix to let GRAS run on top of MSG until we move it directly on top
157 - Various cleanups to the autotools stuff
158 - Begin to move Gras examples to examples/gras/
159 - Let make distcheck work again (yeah!)
161 - documentation overhauled using doxygen.
162 gtk-doc-tools is dead in SimGrid now.
163 - Automatically extract all existing logging categories, and add the list
164 to the documentation (long standing one, to say the less)
166 - Cleanup the known architecture table. Reorder the entries to group what
167 should be, and use a more consistent naming scheme.
168 (some of the test dataset are still to be regenerated)
169 - New! Allow library to register globals on each process just as userdata
171 This is implemented using a xbt_dict and not a xbt_set, so we loose the
172 lookup time (for now).
173 Use it in msg and trp.
174 This cleans a lot the internals and helps enforcing privacy of the
175 headers between the gras components.
176 - New! Add a timer mechanism, not unlike cron(8) and at(1).
177 - Bugfix: gras_os_time was delirious in RL.
178 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
180 - Reenable GRAS now that it works.
182 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
184 SimGrid (2.90) unstable; urgency=low
186 Alpha 1 on the path to SimGrid 3
188 * It is a long time since the last release of SimGrid. I'm sorry about
189 that but as I had told you, I was rewriting a lot of things. I apologize
190 to those who had been reporting bugs to me and that I had not answered.
191 If your bug is still in the new version, please tell me. Here is a
192 summary of the main changes.
194 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
195 lead by Martin Quinson. As a consequence SimGrid gains a lot in
196 portability, speed, and a lot more but you'll figure it out later.
197 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
198 I wanted to release the new MSG as soon as possible and I have
199 broken GRAS, which is the reason why, for now, only MSG is fully
200 functional. A laconic description of these projects is available
201 in the documentation.
203 * REVOLUTION 2: I have removed SG and I am now using a new simulation
204 kernel optimized for our needs (called SURF but only the developers
205 should use it). Hence, MSG is now roughly 30 times faster and I think
206 that by rewriting a little bit MSG, I could event speed it up a little
207 bit more. Beside the gain in speed, it is also much easier to encode a
208 new platform model with SURF than it was with SG. More to come...
210 * REVOLUTION 3: I have tried to change a little as possible the API of
211 MSG but a few things really had to disappear. The main differences
212 with the previous version are :
213 1) no more m_links_t and the corresponding functions. Platforms are
214 directly read from a XML description and cannot be hard-coded
215 anymore. The same format is used for application deployment
216 description. The new format is described in the documentation.
217 Have a look in tools/platform_generation. There is a tiny script
218 that converts from the old platform format to the new one. Concerning
219 the application deployment format, parsing the old one is tricky.
220 I think most of you should however be able to convert your files. If
221 it is really an issue, I can write a C code that does the conversion.
223 2) the toolbox tbx does not exist anymore. We now have a library
224 with much more data-structures but without the hash-tables (we have
225 dictionaries that are much faster).
227 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
229 *****************************************************************************
230 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
231 * don't feel like converting the oldies. (Mt) *
232 *****************************************************************************
235 Version 2.90: "the long awaited one"
236 - Finished rewriting and debugging MSG. Rewrote the documentation.
237 - disable GRAS for now since it needs to be ported to the newest SG
240 - Finish the port to windows (using mingw32 for cross-compile)
243 - Main loop and datastructures of SURF. A cpu resource object is
244 functional. Surf can thus be used to create cpu's with variable
245 performance on which you can execute some actions.
247 2004-11-15 Martin Quinson
248 - Port to ARM. Simply added the alignment and size descriptions. Should
249 work, but the ARM machines are so slow that I didn't had the opportunity
250 to 'make check' over there yet.
252 2004-11-15 Arnaud Legrand
253 - Trace manager now written. It uses a heap structure and is therefore
254 expected to be efficient. It may however be speeded up (particularly
255 when many events occur at the same date) by using red and black
256 trees. One day maybe...
257 - Max-min linear system solver written. It uses a sparse matrix
258 structure taking advantage of its expected use. Most operations are
259 O(1) and free/calloc are called as few as possible. The computation of
260 the minimum could however be improved by using a red and black tree
263 2004-11-03 Arnaud Legrand
264 - Rename every gras_* function that was in xbt/ to its xbt_
266 - Add a heap and a doubly-linked list to xbt
267 - Added a dichotomy to the dictionaries. make check works as well before
268 so I assume that the patch is correct. I do not know however if things
269 run effectively faster than before now. :)
271 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
272 SimGrid, and the version number is bumped to 2.x
274 2004-10-29 Martin Quinson
275 - Introduction of the remote errors.
276 They are the result of a RMI/RPC on the remote machine.
277 ErrCodes being scalar values, you can't get the host on which those
278 errors did happen. Extending the error mechanism as in Gnome is possible.
279 No idea yet whether it is a good idea.
281 2004-10-28 Martin Quinson
282 - Interface revolution: the Starred Structure Eradication.
283 I used to do typedef struct {} toto_t; and then handle *toto_t.
284 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
285 - ???_t is a valid type (builded with typedef)
286 - s_toto_t is a structure (access to fields with .)
287 - s_toto is a structure needing 'struct' keyword to be used
288 - e_toto_t is an enum
289 - toto_t is an 'object' (struct*)
291 typedef struct s_toto {} s_toto_t, *toto_t;
292 typedef enum {} e_toto_t;
293 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
294 s_toto_t) is private.
296 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
297 so that it can set the variable to NULL. It was so for dicts and sets,
298 it changed for dynars.
300 - Fix a bunch of memleaks in dict_remove
301 - Fix a bug in sg/server_socket opening: it failed all the time.
303 2004-10-07 Martin Quinson
304 - Speed up dynar lookup operation a bit.
306 gras_dynar_get is dead.
308 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
309 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
310 (which gives you the address of the stored data).
312 gras_dynar_get_as is an helpful macro which allows you to retrieve a
313 copy of the data using an affectation to do the job and not a memcpy.
315 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
316 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
318 It does not really speedup the dynar test because they are
319 setting elements all the time (and look them seldom). But the dict does
320 far more lookup than setting.
322 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
324 2004-10-05 Martin Quinson
325 - Allow to (en/dis)able the cycle detection at run time.
327 Whether we should check for cycle or not is now a property of each
328 datatype. When you think there may be some cycle, use datadesc_cycle_set.
329 datadesc_cycle_unset allow to remove this property when previously set.
331 Note that the cycle detection is off by default since it impacts the
332 performance. Watch the data you feed GRAS with ;)
334 This property is hereditary. Any element embedded in a structure having it
335 set have it set for the time of this data exchange.
337 You should set it both on sender and receiver side. If you don't set it on
338 sender side, it will enter an endless loop. If you forget on receiver
339 side, the cycles won't be recreated after communication.
341 - Header reorganization.
342 Kill gras_private.h, each submodule must load the headers it needs.
344 2004-10-04 Martin Quinson
345 - Interface revolution: do not try to survive to malloc failure.
347 Now, gras_malloc and friends call gras_abort() on failure.
348 As a conclusion, malloc_error is not a valid error anymore, and all
349 functions for which it was the only gras_error_t return value are
350 changed. They now return void, or there result directly.
351 This simplify the API a lot.
353 2004-09-29 Martin Quinson
354 - Re-enable raw sockets.
355 Created by gras_socket_{client,server}_ext;
356 Used with gras_raw_{send,recv}
359 It should allow to kill the last bits of gras first version soon.
361 This is not completely satisfactory yet (duplicate code with
362 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
365 - Simplify transport plugin (internal) interface by not passing any
366 argument to _server and _client, but embedding them in the socket
369 2004-09-28 Martin Quinson
370 - Finish the port to AIX.
371 autoconf was my problem (segfault within the malloc replacement
372 function. No idea why)
374 2004-09-16 Martin Quinson
375 - Fix some size_t madness on 64bit architectures.
377 2004-09-08 Martin Quinson
378 - Reduce the number of system headers loaded, overload some more system
379 calls (such as malloc to cast the result of the system one, and work
381 - Fix and reintroduce the config support
383 2004-09-07 Martin Quinson
384 - Source code reorganization to allow Arnaud to surf all over there.
385 - Allow to document the logging categories.
386 - Remove all uppercase from logging categories and useless cleanup in names.
388 2004-08-18 Martin Quinson
389 Version 0.6.2 (protocol not changed; API changed)
390 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
391 gras_error_t), and NULL when not found. Functions expecting a msgtype
392 as argument (msg_wait; msg_send) deal with NULL argument by providing a
393 hopefully usefull message.
394 - Portability to prehistoric sparcs again
396 2004-08-17 Martin Quinson
397 Version 0.6.1 (protocol not changed; ABI not changed)
398 - prealloc some buffers to speed things up
400 2004-08-11 Martin Quinson
401 Version 0.6 (protocol not changed; ABI expended)
402 - The parsing macro can deal with the references, provided that you add
403 the relevant annotations (using GRAS_ANNOTE(size,field_name))
405 2004-08-09 Martin Quinson
406 Version 0.5 (protocol not changed; ABI changed)
407 - Allow to off turn the cycle detection code in data exchange at
408 compilation time. It should be at run time, but I'm short of time (and
409 the config stuff is still broken). That way, we keep dict out of the
410 critical path, which is good because the performance is poor:
411 - search not dichotomial yet
412 - dynar give no way to access their content and memcpy everytime
413 - In composed data description (struct, ref and so on), stop foolness of
414 keeping the subtype's ID, but store the type itself. This keeps sets out
415 of the critical path, which is good since they rely on dynar and
416 dictionnaries. The only loose of that is that we cannot detect the
417 redeclaration of a structure/union with another content (but I'm not sure
418 the code detected well this error before anyway). We still can detect
419 the redefinition discrepancy for the other types.
420 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
421 it breaks the code because of type-punning used all over the place).
422 This breaks on all non-gcc architectures (for now).
424 All those changes (plus the buffer of last time) allow me to gain 2 order
425 of magnitude on cruel tests consisting of 800000 array of integers on two
426 level of a hierarchical structure (200 secondes -> 4 secondes)
429 - the selector of reference must now return the type it points to, not
432 2004-08-06 Martin Quinson
433 Version 0.4 (protocol changed; ABI not changed)
434 - Allow to pass --gras-log argument to processes in simulation mode. Really.
435 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
436 - Implement a buffer transport, and use it by default (it relies on tcp in
437 real life and on sg in simulation).
438 That's a bit hackish since I had a new field to the structure to store
439 its data without interfering with the subtype ones. Inheritance
440 is tricky in C. And that's a kind of reverse inheritance with one class
441 derivating two classes. Or maybe a game with java interfaces. Anyway,
442 that's damn hard in C (at least).
443 Moreover, I got tired while trying to ensure plugin separation and
444 genericity in SG mode. MSG wants me to do weird things, so let's go for
445 cruel hacks (temporarily of course ;).
446 See comment in transport_private.h:71
447 - do not include all the _interface headers in private but in the files
448 which really need them (to cut the compilation time when they are
451 2004-07-26 Martin Quinson
452 Version 0.3 (protocol not changed; ABI changed)
453 - Major overhault of the datadesc interface to simplify it:
454 - shorted the function names:
455 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
456 - add a trivial way to push/pop integers into the cbps without malloc.
457 This allows to make really generic sub_type description, which simply
458 pop their size of the stack.
459 - add a function gras_datadesc_ref_pop_arr() which does what users want
460 most of the time: Declare a dynamic array (which pops its size of the
461 stack) and declare a reference to it. Poor name, but anyway.
462 - kill the post-send callback, add a post-receive one
464 2004-07-23 Martin Quinson
465 Version 0.2 (protocol changed; ABI changed)
466 - add some testing for cpbs in the test cases, and fix some more bugs.
467 This invalidate again the little64 data file, since I cannot regenerate
469 - remove an awfull optimization in the logging stuff, allowing me to:
470 - understand it again
471 - learn gcc how to check that the argument match the provided format
472 - fix all errors revealed by gcc after that
473 - internal keys of dict are not \0 terminated. Deal with it properly in
474 loggings instead of segfaulting when the user want to see the logs :-/
476 2004-07-22 Martin Quinson
477 - Fix some stupid bug preventing cbps (callback postit) from working
479 2004-07-21 Martin Quinson
480 - Some documentation cleanups
481 - remove the useless last argument of msgtype_declare
482 - rename the Virtu functions to fit into the 'os' namespace
483 - move headers src/include -> src/include/gras/ and stop fooling with
484 gras -> . symbolic link
485 - make distcheck is now successful
487 2004-07-19 Martin Quinson
489 - Build shared library also
490 - Install html doc to the right location
491 - stop removing maintainer files in make clean
492 - build tests only on make check
494 2004-07-13 Martin Quinson
496 - No major issue in previous version => change versionning schema
497 - Re-enable little64 convertion test now that Abdou kindly regenerated the
498 corresponding dataset.
500 2004-07-11 Martin Quinson
502 - Get it working with any kind of structure (we can compute the padding
503 bytes remotely for all the architectures I have access to)
504 - Implement the structure parsing macro (still not quite robust/complete)
505 - Improvement to the remote testing toysuite
507 2004-07-10 Martin Quinson
509 - get ride of a bunch of deprecated macros
510 - actually run the test for two-compliment, not only compile it :-/
511 - test whether the structures get packed (and bail out if yes. Damn.
512 Alignment is a serious matter)
513 - test whether the structures get compacted (but respecting the alignment
514 constraints of each types)
515 - test whether the array fields of structures can straddle alignment boundaries
517 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
518 (took me 2 hours to find that bug, looking at the wrong place)
519 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
520 before sending/receiving and are used to compute the offsets of fields
521 - Given that padding size depend even on compiler options, keep track of
522 alignment and aligned_size only for the current architecture. Not a big
523 deal since we send structure fields one after the other (seems
525 - Add the datastructure used for IEEE paper by the PBIO guys to the test
526 program, let it work on linux/gcc/little32. portability todo.
528 2004-07-08 Martin Quinson
529 - import and improve remote compilation support from FAST
530 - make sure make check works on half a dozen of machines out there
532 2004-07-07 Martin Quinson
533 Let's say it's version 0.0.3 ;)
534 - Implement conversions (yuhu!)
535 - Let it work on solaris (beside conversion, of course)
536 - Stupid me, using rand() to generate the conversion datatests in not wise.
538 2004-07-06 Martin Quinson
539 - Let make dist work, since I'm gonna need it to compile on remote hosts
540 - Let Tests/datadesc_usage write the architecture on which the file was
541 generated as first byte.
542 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
543 architecture descriptions.
544 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
545 on those architectures.
546 - Optimization: send/recv array of scalar in one shoot
548 2004-07-05 Martin Quinson
549 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
551 - Plug a whole bunch of memleaks
552 - each process now have to call gras_{init,exit}. One day, their log
553 settings will be separated
554 - Continue the code factorisation between SG, RL and common in Transport.
556 2004-07-04 Martin Quinson
558 - Redistribution between SG and RL.
559 We wanna have to accept in SG, so move accepted related parts of RL in
560 the common part. (more precisely, the dynar of all known sockets is no
561 more a static in transport.c, but part of the process_data)
563 [gras_stub_generator]
564 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
566 2004-07-03 Martin Quinson
567 - Create a new log channel tbx containing dict, set, log, dynar (to shut
568 them all up in one shot)
570 - Fix the ugly case of reference to dynamic array.
571 - New (semi-public) function gras_datadesc_size to allow the messaging
572 layer to malloc the needed space for the buffer.
574 - gras_socket_close now expect the socket to close (and not its address to
575 put NULL in it after it). This is because the socket passed to handlers
576 is one of their argument (=> not writable).
578 - propagate the interface cleanup from last week in datadesc, ie remove a
579 superfluous level of indirection. User pass adress of variable
580 containing data (both when sending and receiving), and not of a variable
581 being a pointer to the data. Let's say that I like it better ;)
582 The price for that is constructs like "int msg=*(int*)payload" in
583 handlers, but it's a fine price, IMHO.
585 - Let it work in RL (yuhu)
587 2004-06-21 Martin Quinson
589 - porting SG plugin and SG select to new standards (works almost).
590 - plug memleaks and fix bugs around.
593 - cleanup the prototype of data recv and force users to specify when they
594 want to handle references to objects. Test case working even for cycles.
595 - plug memleaks. Valgrind is perfectly ok with this.
597 2004-06-12 Martin Quinson
599 - cleanup the separation between plugin and main code in plugin creation
601 2004-06-11 Martin Quinson
603 - Reput hook for raw sockets, needed for BW experiments
604 - kill a few lines of dead code
605 [Data description] Interface cleanup
606 - gras_datadesc_by_name returns the searched type or NULL.
607 That way, no variable is needed to use a type desc once, which makes
609 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
610 two parameters were strings (field name, type name), leading to
612 [Dicos] Interface cleanup
613 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
614 This is consistant with the dynar API.
616 2004-04-21 Martin Quinson
618 - Porting to new standards.
621 There is no bag anymore, no need to take extra provision to mask the
622 pointers behind "ID".
623 Better splitup of functions between files create/exchange/convert.
624 This is still a bit artificial since convert and receive are so
625 interleaved, but anyway.
627 - add a queued message list to procdata (the ones not matching criteria
629 - factorize some more code between SG and RL wrt procdata
631 - use gras_exit in example to track memleaks
632 - get rid of gs_example now that GS is properly integrated into gras
633 - update run_test to integrate the lastest tests (datadesc)
635 - rename WARNINGn macros to WARNn since it prooved error-prone
637 2004-04-19 Martin Quinson
639 - register init/exit functions within gras module mechanism
640 - send/receive function.
641 Convertion is not implemented, but short-cutted if not needed.
642 struct/array elements are sent one by one (instead of block-wise), but
643 nobody really cares (yet). Get a prototype before optimizing.
644 - tests (using a file socket) for DD send/receive on:
645 - base types: int, float
646 - array: fixed size, string (ie ref to dynamic string)
647 - structure: homogeneous, heterogeneous
648 - chained list, graph with cycle
649 Believe it or not, valgrind is not too unhappy with the results. The
650 cycle happily segfaults, but the others are ok. And I'm sick of pointers
654 - Bugfix when using a filename explicitely (instead of '-')
656 2004-04-09 Martin Quinson
658 - factorize more code between RL and SG in socket creation
659 - Complete the implementation and tests of:
661 o file (only in RL, and mainly for debugging)
663 I lost 3 days to design a portable address resolver, and then decided
664 that the prototype mainly have to run on my box.
665 Addressing portability too early may be like optimizing too early :-/
667 - use gras_init in the Tests instead of the crappy parse_log_opt
668 (the latter function is removed)
669 [Conditional execution]
670 - New functions: gras_if_RL/gras_if_SG (basic support for this)
671 [Code reorganisation]
672 - Get rid of libgrasutils.a since it makes more trouble than it solves.
673 Build examples against the RL library, since there is no way to disable
674 its creation for now.