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 * Bugfix: Allow absolute paths for platform description files [MQ]
10 * Bugfix: do free the variables after use. Leads to drastic performance
12 * Implement max_duration (ie, timeouts) on resources [AL]
15 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
17 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
18 the needed support (timeouts) [AL]
21 * Implement measurement sockets. You can now get the bandwidth between two
22 hosts thanks to AMOK (see below). [MQ]
23 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
24 dynar over the network (yeah) [MQ]
25 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
26 * simple/static token-ring example. [Alexandre Colucci and MQ]
27 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
31 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
32 type parsing. "short" and "long" modifiers were also ignored; other
33 modifier (such as reference level) are still ignored. [MQ]
34 * Embeed the buffer size within the buffer itself on SG. [MQ]
35 That way, send() are atomic and cannot get intermixed anymore (at least
36 the ones which are less than 100k; bigger messages still have the issue)
37 * Array size pushed by the field, not by the field type (or each
38 and every long int will push stuff to the cbps) [MQ]
39 * use select() to sleep since it allows to portably sleep less than one
43 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
46 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
47 * type category 'ignored' killed as it was never used and were difficult
49 * whether a type can cycle or not is now a flag, leaving room for more
50 flags (as "ignored", if we feel the need one day ;)
51 * Rename raw sockets to measurement sockets since "raw" has another
52 meaning in networking community.
55 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
56 and offers features not belonging to GRAS but that most applications
57 need. One day, it may be a set of plugins loadable at runtime.
58 * New module: bandwidth
59 bandwidth measurement between arbitrary nodes running this module. [MQ]
63 SimGrid (2.94) unstable; urgency=low
65 The first beta release of SimGrid 3 !
69 * Update the main page and the FAQ. Adding references to gforge.
72 * Add a gras_os_getpid function.
75 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
76 * Extend the logs so that they also print PID, hostname, date, ... if
78 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
79 and kill the old version which were in testsuite/
80 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
84 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
85 for users. Instead, each programming environment (GRAS, MSG,...) use it
86 when needed to provide such a feature to users.
87 Don't shortcut the mecanism or you will also shortcut the virtualization
88 you need on the simulator.
92 * Cleanups in configury with regard to compile optimization/warning flags.
93 Also add -fno-loop-optimize to any powerpc since it's the optimization
94 killing gcc (< 3.4.0).
95 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
96 needed by MSG examples complications
97 * Borrow configury beautifications from PHP
100 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
101 warning (thanks loris for stumbling into this one).
102 * Bugfix: stop loading private headers (gras_config.h) from the public
106 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
107 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
108 RL would imply a DNS resolver, which is *hard* to do in a portable way
109 (and therefore delayed).
110 * Implement a real timer mecanism and use it in timing macros. This allows
111 to avoid rounding errors and get a 0.000005 sec precision in timing
112 macros. While I was at it, various cleanups:
113 - allow to declare more than one timed section per file (fix a stupid bug)
114 - move some private declaration to the right place
115 - merge conditional execution and timing macros into emulation module
116 - document the module
117 - make sure the module cleanups its mess on gras_exit
118 * Documentation improvements:
119 - (new) how to compile applications using GRAS
120 - (new) emulation support (timing macros)
122 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
124 SimGrid (2.93) unstable; urgency=low
126 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
129 - Use Paje properly where used. Still to be sanitized properly.
130 - Portability fix: Add an implementation of the contexts using pthread
134 - Add xbt_procname(): returns the name of the current process.
135 Use it to show the current process's name in all logging.
137 - fix detection of older flex version and the reaction, since we do
138 depend on modern ones (we use lex_destroy)
139 - Better separation of SG and RL in the libs: remove all simulation code
140 from libgras. As a result, this lib is now only 200k when stripped.
141 Some of the xbt modules may also be duplicated (two sets and such) and
142 should be cleaned/killed before SG3.
143 - Insist on using xlC on AIX because of weird problems involving gcc there.
144 - Cleanup the make remote stuff. This is now done by scripts
145 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
146 mainly for our private use, but we're working on changing them to user
149 - Bugfix: flush the socket on close only if there is some *output*.
150 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
152 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
155 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
157 SimGrid (2.92) unstable; urgency=low
159 Alpha 3 on the path to SimGrid 3
163 - New! First try of benchmarking macros.
164 - New! First try so that gras_stub_generator generate deployment and
165 remote compilation helpers.
167 - Bugfix: Initialization fix in msg_test.
171 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
173 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
175 - API Change: gras_os_sleep to take the amount of seconds as a double.
176 Accepting an int was error prone since it was the only location where
177 seconds were coded as such. It leaded to damn rounding errors.
178 - Bugfix: Hard to belive that timers ever worked before this.
180 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
182 SimGrid (2.91) unstable; urgency=low
184 Alpha 2 on the path to SimGrid 3
188 - Bug fix in the lmm_solver.
190 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
191 through the function MSG_paje_output.
192 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
193 - It is possible to bound the rate of a communication in MSG with
194 MSG_task_put_bounded() (was already in the previous version but I had forgotten
195 to write it in the changelog).
196 - Bug fix to let GRAS run on top of MSG until we move it directly on top
201 - Various cleanups to the autotools stuff
202 - Begin to move Gras examples to examples/gras/
203 - Let make distcheck work again (yeah!)
205 - documentation overhauled using doxygen.
206 gtk-doc-tools is dead in SimGrid now.
207 - Automatically extract all existing logging categories, and add the list
208 to the documentation (long standing one, to say the less)
210 - Cleanup the known architecture table. Reorder the entries to group what
211 should be, and use a more consistent naming scheme.
212 (some of the test dataset are still to be regenerated)
213 - New! Allow library to register globals on each process just as userdata
215 This is implemented using a xbt_dict and not a xbt_set, so we loose the
216 lookup time (for now).
217 Use it in msg and trp.
218 This cleans a lot the internals and helps enforcing privacy of the
219 headers between the gras components.
220 - New! Add a timer mechanism, not unlike cron(8) and at(1).
221 - Bugfix: gras_os_time was delirious in RL.
222 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
224 - Reenable GRAS now that it works.
226 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
228 SimGrid (2.90) unstable; urgency=low
230 Alpha 1 on the path to SimGrid 3
232 * It is a long time since the last release of SimGrid. I'm sorry about
233 that but as I had told you, I was rewriting a lot of things. I apologize
234 to those who had been reporting bugs to me and that I had not answered.
235 If your bug is still in the new version, please tell me. Here is a
236 summary of the main changes.
238 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
239 lead by Martin Quinson. As a consequence SimGrid gains a lot in
240 portability, speed, and a lot more but you'll figure it out later.
241 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
242 I wanted to release the new MSG as soon as possible and I have
243 broken GRAS, which is the reason why, for now, only MSG is fully
244 functional. A laconic description of these projects is available
245 in the documentation.
247 * REVOLUTION 2: I have removed SG and I am now using a new simulation
248 kernel optimized for our needs (called SURF but only the developers
249 should use it). Hence, MSG is now roughly 30 times faster and I think
250 that by rewriting a little bit MSG, I could event speed it up a little
251 bit more. Beside the gain in speed, it is also much easier to encode a
252 new platform model with SURF than it was with SG. More to come...
254 * REVOLUTION 3: I have tried to change a little as possible the API of
255 MSG but a few things really had to disappear. The main differences
256 with the previous version are :
257 1) no more m_links_t and the corresponding functions. Platforms are
258 directly read from a XML description and cannot be hard-coded
259 anymore. The same format is used for application deployment
260 description. The new format is described in the documentation.
261 Have a look in tools/platform_generation. There is a tiny script
262 that converts from the old platform format to the new one. Concerning
263 the application deployment format, parsing the old one is tricky.
264 I think most of you should however be able to convert your files. If
265 it is really an issue, I can write a C code that does the conversion.
267 2) the toolbox tbx does not exist anymore. We now have a library
268 with much more data-structures but without the hash-tables (we have
269 dictionaries that are much faster).
271 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
273 *****************************************************************************
274 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
275 * don't feel like converting the oldies. (Mt) *
276 *****************************************************************************
279 Version 2.90: "the long awaited one"
280 - Finished rewriting and debugging MSG. Rewrote the documentation.
281 - disable GRAS for now since it needs to be ported to the newest SG
284 - Finish the port to windows (using mingw32 for cross-compile)
287 - Main loop and datastructures of SURF. A cpu resource object is
288 functional. Surf can thus be used to create cpu's with variable
289 performance on which you can execute some actions.
291 2004-11-15 Martin Quinson
292 - Port to ARM. Simply added the alignment and size descriptions. Should
293 work, but the ARM machines are so slow that I didn't had the opportunity
294 to 'make check' over there yet.
296 2004-11-15 Arnaud Legrand
297 - Trace manager now written. It uses a heap structure and is therefore
298 expected to be efficient. It may however be speeded up (particularly
299 when many events occur at the same date) by using red and black
300 trees. One day maybe...
301 - Max-min linear system solver written. It uses a sparse matrix
302 structure taking advantage of its expected use. Most operations are
303 O(1) and free/calloc are called as few as possible. The computation of
304 the minimum could however be improved by using a red and black tree
307 2004-11-03 Arnaud Legrand
308 - Rename every gras_* function that was in xbt/ to its xbt_
310 - Add a heap and a doubly-linked list to xbt
311 - Added a dichotomy to the dictionaries. make check works as well before
312 so I assume that the patch is correct. I do not know however if things
313 run effectively faster than before now. :)
315 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
316 SimGrid, and the version number is bumped to 2.x
318 2004-10-29 Martin Quinson
319 - Introduction of the remote errors.
320 They are the result of a RMI/RPC on the remote machine.
321 ErrCodes being scalar values, you can't get the host on which those
322 errors did happen. Extending the error mechanism as in Gnome is possible.
323 No idea yet whether it is a good idea.
325 2004-10-28 Martin Quinson
326 - Interface revolution: the Starred Structure Eradication.
327 I used to do typedef struct {} toto_t; and then handle *toto_t.
328 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
329 - ???_t is a valid type (builded with typedef)
330 - s_toto_t is a structure (access to fields with .)
331 - s_toto is a structure needing 'struct' keyword to be used
332 - e_toto_t is an enum
333 - toto_t is an 'object' (struct*)
335 typedef struct s_toto {} s_toto_t, *toto_t;
336 typedef enum {} e_toto_t;
337 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
338 s_toto_t) is private.
340 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
341 so that it can set the variable to NULL. It was so for dicts and sets,
342 it changed for dynars.
344 - Fix a bunch of memleaks in dict_remove
345 - Fix a bug in sg/server_socket opening: it failed all the time.
347 2004-10-07 Martin Quinson
348 - Speed up dynar lookup operation a bit.
350 gras_dynar_get is dead.
352 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
353 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
354 (which gives you the address of the stored data).
356 gras_dynar_get_as is an helpful macro which allows you to retrieve a
357 copy of the data using an affectation to do the job and not a memcpy.
359 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
360 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
362 It does not really speedup the dynar test because they are
363 setting elements all the time (and look them seldom). But the dict does
364 far more lookup than setting.
366 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
368 2004-10-05 Martin Quinson
369 - Allow to (en/dis)able the cycle detection at run time.
371 Whether we should check for cycle or not is now a property of each
372 datatype. When you think there may be some cycle, use datadesc_cycle_set.
373 datadesc_cycle_unset allow to remove this property when previously set.
375 Note that the cycle detection is off by default since it impacts the
376 performance. Watch the data you feed GRAS with ;)
378 This property is hereditary. Any element embedded in a structure having it
379 set have it set for the time of this data exchange.
381 You should set it both on sender and receiver side. If you don't set it on
382 sender side, it will enter an endless loop. If you forget on receiver
383 side, the cycles won't be recreated after communication.
385 - Header reorganization.
386 Kill gras_private.h, each submodule must load the headers it needs.
388 2004-10-04 Martin Quinson
389 - Interface revolution: do not try to survive to malloc failure.
391 Now, gras_malloc and friends call gras_abort() on failure.
392 As a conclusion, malloc_error is not a valid error anymore, and all
393 functions for which it was the only gras_error_t return value are
394 changed. They now return void, or there result directly.
395 This simplify the API a lot.
397 2004-09-29 Martin Quinson
398 - Re-enable raw sockets.
399 Created by gras_socket_{client,server}_ext;
400 Used with gras_raw_{send,recv}
403 It should allow to kill the last bits of gras first version soon.
405 This is not completely satisfactory yet (duplicate code with
406 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
409 - Simplify transport plugin (internal) interface by not passing any
410 argument to _server and _client, but embedding them in the socket
413 2004-09-28 Martin Quinson
414 - Finish the port to AIX.
415 autoconf was my problem (segfault within the malloc replacement
416 function. No idea why)
418 2004-09-16 Martin Quinson
419 - Fix some size_t madness on 64bit architectures.
421 2004-09-08 Martin Quinson
422 - Reduce the number of system headers loaded, overload some more system
423 calls (such as malloc to cast the result of the system one, and work
425 - Fix and reintroduce the config support
427 2004-09-07 Martin Quinson
428 - Source code reorganization to allow Arnaud to surf all over there.
429 - Allow to document the logging categories.
430 - Remove all uppercase from logging categories and useless cleanup in names.
432 2004-08-18 Martin Quinson
433 Version 0.6.2 (protocol not changed; API changed)
434 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
435 gras_error_t), and NULL when not found. Functions expecting a msgtype
436 as argument (msg_wait; msg_send) deal with NULL argument by providing a
437 hopefully usefull message.
438 - Portability to prehistoric sparcs again
440 2004-08-17 Martin Quinson
441 Version 0.6.1 (protocol not changed; ABI not changed)
442 - prealloc some buffers to speed things up
444 2004-08-11 Martin Quinson
445 Version 0.6 (protocol not changed; ABI expended)
446 - The parsing macro can deal with the references, provided that you add
447 the relevant annotations (using GRAS_ANNOTE(size,field_name))
449 2004-08-09 Martin Quinson
450 Version 0.5 (protocol not changed; ABI changed)
451 - Allow to off turn the cycle detection code in data exchange at
452 compilation time. It should be at run time, but I'm short of time (and
453 the config stuff is still broken). That way, we keep dict out of the
454 critical path, which is good because the performance is poor:
455 - search not dichotomial yet
456 - dynar give no way to access their content and memcpy everytime
457 - In composed data description (struct, ref and so on), stop foolness of
458 keeping the subtype's ID, but store the type itself. This keeps sets out
459 of the critical path, which is good since they rely on dynar and
460 dictionnaries. The only loose of that is that we cannot detect the
461 redeclaration of a structure/union with another content (but I'm not sure
462 the code detected well this error before anyway). We still can detect
463 the redefinition discrepancy for the other types.
464 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
465 it breaks the code because of type-punning used all over the place).
466 This breaks on all non-gcc architectures (for now).
468 All those changes (plus the buffer of last time) allow me to gain 2 order
469 of magnitude on cruel tests consisting of 800000 array of integers on two
470 level of a hierarchical structure (200 secondes -> 4 secondes)
473 - the selector of reference must now return the type it points to, not
476 2004-08-06 Martin Quinson
477 Version 0.4 (protocol changed; ABI not changed)
478 - Allow to pass --gras-log argument to processes in simulation mode. Really.
479 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
480 - Implement a buffer transport, and use it by default (it relies on tcp in
481 real life and on sg in simulation).
482 That's a bit hackish since I had a new field to the structure to store
483 its data without interfering with the subtype ones. Inheritance
484 is tricky in C. And that's a kind of reverse inheritance with one class
485 derivating two classes. Or maybe a game with java interfaces. Anyway,
486 that's damn hard in C (at least).
487 Moreover, I got tired while trying to ensure plugin separation and
488 genericity in SG mode. MSG wants me to do weird things, so let's go for
489 cruel hacks (temporarily of course ;).
490 See comment in transport_private.h:71
491 - do not include all the _interface headers in private but in the files
492 which really need them (to cut the compilation time when they are
495 2004-07-26 Martin Quinson
496 Version 0.3 (protocol not changed; ABI changed)
497 - Major overhault of the datadesc interface to simplify it:
498 - shorted the function names:
499 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
500 - add a trivial way to push/pop integers into the cbps without malloc.
501 This allows to make really generic sub_type description, which simply
502 pop their size of the stack.
503 - add a function gras_datadesc_ref_pop_arr() which does what users want
504 most of the time: Declare a dynamic array (which pops its size of the
505 stack) and declare a reference to it. Poor name, but anyway.
506 - kill the post-send callback, add a post-receive one
508 2004-07-23 Martin Quinson
509 Version 0.2 (protocol changed; ABI changed)
510 - add some testing for cpbs in the test cases, and fix some more bugs.
511 This invalidate again the little64 data file, since I cannot regenerate
513 - remove an awfull optimization in the logging stuff, allowing me to:
514 - understand it again
515 - learn gcc how to check that the argument match the provided format
516 - fix all errors revealed by gcc after that
517 - internal keys of dict are not \0 terminated. Deal with it properly in
518 loggings instead of segfaulting when the user want to see the logs :-/
520 2004-07-22 Martin Quinson
521 - Fix some stupid bug preventing cbps (callback postit) from working
523 2004-07-21 Martin Quinson
524 - Some documentation cleanups
525 - remove the useless last argument of msgtype_declare
526 - rename the Virtu functions to fit into the 'os' namespace
527 - move headers src/include -> src/include/gras/ and stop fooling with
528 gras -> . symbolic link
529 - make distcheck is now successful
531 2004-07-19 Martin Quinson
533 - Build shared library also
534 - Install html doc to the right location
535 - stop removing maintainer files in make clean
536 - build tests only on make check
538 2004-07-13 Martin Quinson
540 - No major issue in previous version => change versionning schema
541 - Re-enable little64 convertion test now that Abdou kindly regenerated the
542 corresponding dataset.
544 2004-07-11 Martin Quinson
546 - Get it working with any kind of structure (we can compute the padding
547 bytes remotely for all the architectures I have access to)
548 - Implement the structure parsing macro (still not quite robust/complete)
549 - Improvement to the remote testing toysuite
551 2004-07-10 Martin Quinson
553 - get ride of a bunch of deprecated macros
554 - actually run the test for two-compliment, not only compile it :-/
555 - test whether the structures get packed (and bail out if yes. Damn.
556 Alignment is a serious matter)
557 - test whether the structures get compacted (but respecting the alignment
558 constraints of each types)
559 - test whether the array fields of structures can straddle alignment boundaries
561 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
562 (took me 2 hours to find that bug, looking at the wrong place)
563 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
564 before sending/receiving and are used to compute the offsets of fields
565 - Given that padding size depend even on compiler options, keep track of
566 alignment and aligned_size only for the current architecture. Not a big
567 deal since we send structure fields one after the other (seems
569 - Add the datastructure used for IEEE paper by the PBIO guys to the test
570 program, let it work on linux/gcc/little32. portability todo.
572 2004-07-08 Martin Quinson
573 - import and improve remote compilation support from FAST
574 - make sure make check works on half a dozen of machines out there
576 2004-07-07 Martin Quinson
577 Let's say it's version 0.0.3 ;)
578 - Implement conversions (yuhu!)
579 - Let it work on solaris (beside conversion, of course)
580 - Stupid me, using rand() to generate the conversion datatests in not wise.
582 2004-07-06 Martin Quinson
583 - Let make dist work, since I'm gonna need it to compile on remote hosts
584 - Let Tests/datadesc_usage write the architecture on which the file was
585 generated as first byte.
586 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
587 architecture descriptions.
588 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
589 on those architectures.
590 - Optimization: send/recv array of scalar in one shoot
592 2004-07-05 Martin Quinson
593 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
595 - Plug a whole bunch of memleaks
596 - each process now have to call gras_{init,exit}. One day, their log
597 settings will be separated
598 - Continue the code factorisation between SG, RL and common in Transport.
600 2004-07-04 Martin Quinson
602 - Redistribution between SG and RL.
603 We wanna have to accept in SG, so move accepted related parts of RL in
604 the common part. (more precisely, the dynar of all known sockets is no
605 more a static in transport.c, but part of the process_data)
607 [gras_stub_generator]
608 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
610 2004-07-03 Martin Quinson
611 - Create a new log channel tbx containing dict, set, log, dynar (to shut
612 them all up in one shot)
614 - Fix the ugly case of reference to dynamic array.
615 - New (semi-public) function gras_datadesc_size to allow the messaging
616 layer to malloc the needed space for the buffer.
618 - gras_socket_close now expect the socket to close (and not its address to
619 put NULL in it after it). This is because the socket passed to handlers
620 is one of their argument (=> not writable).
622 - propagate the interface cleanup from last week in datadesc, ie remove a
623 superfluous level of indirection. User pass adress of variable
624 containing data (both when sending and receiving), and not of a variable
625 being a pointer to the data. Let's say that I like it better ;)
626 The price for that is constructs like "int msg=*(int*)payload" in
627 handlers, but it's a fine price, IMHO.
629 - Let it work in RL (yuhu)
631 2004-06-21 Martin Quinson
633 - porting SG plugin and SG select to new standards (works almost).
634 - plug memleaks and fix bugs around.
637 - cleanup the prototype of data recv and force users to specify when they
638 want to handle references to objects. Test case working even for cycles.
639 - plug memleaks. Valgrind is perfectly ok with this.
641 2004-06-12 Martin Quinson
643 - cleanup the separation between plugin and main code in plugin creation
645 2004-06-11 Martin Quinson
647 - Reput hook for raw sockets, needed for BW experiments
648 - kill a few lines of dead code
649 [Data description] Interface cleanup
650 - gras_datadesc_by_name returns the searched type or NULL.
651 That way, no variable is needed to use a type desc once, which makes
653 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
654 two parameters were strings (field name, type name), leading to
656 [Dicos] Interface cleanup
657 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
658 This is consistant with the dynar API.
660 2004-04-21 Martin Quinson
662 - Porting to new standards.
665 There is no bag anymore, no need to take extra provision to mask the
666 pointers behind "ID".
667 Better splitup of functions between files create/exchange/convert.
668 This is still a bit artificial since convert and receive are so
669 interleaved, but anyway.
671 - add a queued message list to procdata (the ones not matching criteria
673 - factorize some more code between SG and RL wrt procdata
675 - use gras_exit in example to track memleaks
676 - get rid of gs_example now that GS is properly integrated into gras
677 - update run_test to integrate the lastest tests (datadesc)
679 - rename WARNINGn macros to WARNn since it prooved error-prone
681 2004-04-19 Martin Quinson
683 - register init/exit functions within gras module mechanism
684 - send/receive function.
685 Convertion is not implemented, but short-cutted if not needed.
686 struct/array elements are sent one by one (instead of block-wise), but
687 nobody really cares (yet). Get a prototype before optimizing.
688 - tests (using a file socket) for DD send/receive on:
689 - base types: int, float
690 - array: fixed size, string (ie ref to dynamic string)
691 - structure: homogeneous, heterogeneous
692 - chained list, graph with cycle
693 Believe it or not, valgrind is not too unhappy with the results. The
694 cycle happily segfaults, but the others are ok. And I'm sick of pointers
698 - Bugfix when using a filename explicitely (instead of '-')
700 2004-04-09 Martin Quinson
702 - factorize more code between RL and SG in socket creation
703 - Complete the implementation and tests of:
705 o file (only in RL, and mainly for debugging)
707 I lost 3 days to design a portable address resolver, and then decided
708 that the prototype mainly have to run on my box.
709 Addressing portability too early may be like optimizing too early :-/
711 - use gras_init in the Tests instead of the crappy parse_log_opt
712 (the latter function is removed)
713 [Conditional execution]
714 - New functions: gras_if_RL/gras_if_SG (basic support for this)
715 [Code reorganisation]
716 - Get rid of libgrasutils.a since it makes more trouble than it solves.
717 Build examples against the RL library, since there is no way to disable
718 its creation for now.