1 SimGrid (2.96) unstable; urgency=low
4 * New! Add a FATPIPE model. [AL]
5 * Naw! Add a parallel task model. [AL]
8 * Bugfix: MSG_process_resume now works with the current running process.
10 * New! Add MSG_parallel_task_create and MSG_parallel_task_execute. [AL]
11 * Modification of MSG_task_get_compute_duration. Once a task has been
12 processed, the value returned by this function is now equal to 0. [AL]
13 * New! Add double MSG_task_get_remaining_computation(m_task_t task) and
14 MSG_error_t MSG_task_cancel(m_task_t task). Add a state to MSG_error_t
15 corresponding to the cancelation of a m_task. For now, MSG_task_cancel
16 only works with computation tasks. [AL]
17 * New! Add double MSG_get_host_speed(m_host_t h) that returns the speed
18 of the processor (in Mflop/s) regardless of the current load on the
23 SimGrid (2.95) unstable; urgency=low
26 * Steal some nice code to GNU pth to fix context detection and usage [AL]
27 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
28 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
31 * Bugfix: Allow absolute paths for platform description files [MQ]
32 * Bugfix: do free the variables after use. Leads to drastic performance
34 * Implement max_duration (ie, timeouts) on resources [AL]
37 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
39 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
40 the needed support (timeouts) [AL]
43 * Implement measurement sockets. You can now get the bandwidth between two
44 hosts thanks to AMOK (see below). [MQ]
45 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
46 dynar over the network (yeah) [MQ]
47 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
48 * simple/static token-ring example. [Alexandre Colucci and MQ]
49 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
53 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
54 type parsing. "short" and "long" modifiers were also ignored; other
55 modifier (such as reference level) are still ignored. [MQ]
56 * Embeed the buffer size within the buffer itself on SG. [MQ]
57 That way, send() are atomic and cannot get intermixed anymore (at least
58 the ones which are less than 100k; bigger messages still have the issue)
59 * Array size pushed by the field, not by the field type (or each
60 and every long int will push stuff to the cbps) [MQ]
61 * use select() to sleep since it allows to portably sleep less than one
65 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
68 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
69 * type category 'ignored' killed as it was never used and were difficult
71 * whether a type can cycle or not is now a flag, leaving room for more
72 flags (as "ignored", if we feel the need one day ;)
73 * Rename raw sockets to measurement sockets since "raw" has another
74 meaning in networking community.
77 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
78 and offers features not belonging to GRAS but that most applications
79 need. One day, it may be a set of plugins loadable at runtime.
80 * New module: bandwidth
81 bandwidth measurement between arbitrary nodes running this module. [MQ]
83 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
85 SimGrid (2.94) unstable; urgency=low
87 The first beta release of SimGrid 3 !
91 * Update the main page and the FAQ. Adding references to gforge.
94 * Add a gras_os_getpid function.
97 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
98 * Extend the logs so that they also print PID, hostname, date, ... if
100 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
101 and kill the old version which were in testsuite/
102 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
106 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
107 for users. Instead, each programming environment (GRAS, MSG,...) use it
108 when needed to provide such a feature to users.
109 Don't shortcut the mecanism or you will also shortcut the virtualization
110 you need on the simulator.
114 * Cleanups in configury with regard to compile optimization/warning flags.
115 Also add -fno-loop-optimize to any powerpc since it's the optimization
116 killing gcc (< 3.4.0).
117 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
118 needed by MSG examples complications
119 * Borrow configury beautifications from PHP
122 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
123 warning (thanks loris for stumbling into this one).
124 * Bugfix: stop loading private headers (gras_config.h) from the public
128 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
129 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
130 RL would imply a DNS resolver, which is *hard* to do in a portable way
131 (and therefore delayed).
132 * Implement a real timer mecanism and use it in timing macros. This allows
133 to avoid rounding errors and get a 0.000005 sec precision in timing
134 macros. While I was at it, various cleanups:
135 - allow to declare more than one timed section per file (fix a stupid bug)
136 - move some private declaration to the right place
137 - merge conditional execution and timing macros into emulation module
138 - document the module
139 - make sure the module cleanups its mess on gras_exit
140 * Documentation improvements:
141 - (new) how to compile applications using GRAS
142 - (new) emulation support (timing macros)
144 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
146 SimGrid (2.93) unstable; urgency=low
148 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
151 - Use Paje properly where used. Still to be sanitized properly.
152 - Portability fix: Add an implementation of the contexts using pthread
156 - Add xbt_procname(): returns the name of the current process.
157 Use it to show the current process's name in all logging.
159 - fix detection of older flex version and the reaction, since we do
160 depend on modern ones (we use lex_destroy)
161 - Better separation of SG and RL in the libs: remove all simulation code
162 from libgras. As a result, this lib is now only 200k when stripped.
163 Some of the xbt modules may also be duplicated (two sets and such) and
164 should be cleaned/killed before SG3.
165 - Insist on using xlC on AIX because of weird problems involving gcc there.
166 - Cleanup the make remote stuff. This is now done by scripts
167 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
168 mainly for our private use, but we're working on changing them to user
171 - Bugfix: flush the socket on close only if there is some *output*.
172 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
174 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
177 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
179 SimGrid (2.92) unstable; urgency=low
181 Alpha 3 on the path to SimGrid 3
185 - New! First try of benchmarking macros.
186 - New! First try so that gras_stub_generator generate deployment and
187 remote compilation helpers.
189 - Bugfix: Initialization fix in msg_test.
193 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
195 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
197 - API Change: gras_os_sleep to take the amount of seconds as a double.
198 Accepting an int was error prone since it was the only location where
199 seconds were coded as such. It leaded to damn rounding errors.
200 - Bugfix: Hard to belive that timers ever worked before this.
202 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
204 SimGrid (2.91) unstable; urgency=low
206 Alpha 2 on the path to SimGrid 3
210 - Bug fix in the lmm_solver.
212 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
213 through the function MSG_paje_output.
214 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
215 - It is possible to bound the rate of a communication in MSG with
216 MSG_task_put_bounded() (was already in the previous version but I had forgotten
217 to write it in the changelog).
218 - Bug fix to let GRAS run on top of MSG until we move it directly on top
223 - Various cleanups to the autotools stuff
224 - Begin to move Gras examples to examples/gras/
225 - Let make distcheck work again (yeah!)
227 - documentation overhauled using doxygen.
228 gtk-doc-tools is dead in SimGrid now.
229 - Automatically extract all existing logging categories, and add the list
230 to the documentation (long standing one, to say the less)
232 - Cleanup the known architecture table. Reorder the entries to group what
233 should be, and use a more consistent naming scheme.
234 (some of the test dataset are still to be regenerated)
235 - New! Allow library to register globals on each process just as userdata
237 This is implemented using a xbt_dict and not a xbt_set, so we loose the
238 lookup time (for now).
239 Use it in msg and trp.
240 This cleans a lot the internals and helps enforcing privacy of the
241 headers between the gras components.
242 - New! Add a timer mechanism, not unlike cron(8) and at(1).
243 - Bugfix: gras_os_time was delirious in RL.
244 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
246 - Reenable GRAS now that it works.
248 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
250 SimGrid (2.90) unstable; urgency=low
252 Alpha 1 on the path to SimGrid 3
254 * It is a long time since the last release of SimGrid. I'm sorry about
255 that but as I had told you, I was rewriting a lot of things. I apologize
256 to those who had been reporting bugs to me and that I had not answered.
257 If your bug is still in the new version, please tell me. Here is a
258 summary of the main changes.
260 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
261 lead by Martin Quinson. As a consequence SimGrid gains a lot in
262 portability, speed, and a lot more but you'll figure it out later.
263 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
264 I wanted to release the new MSG as soon as possible and I have
265 broken GRAS, which is the reason why, for now, only MSG is fully
266 functional. A laconic description of these projects is available
267 in the documentation.
269 * REVOLUTION 2: I have removed SG and I am now using a new simulation
270 kernel optimized for our needs (called SURF but only the developers
271 should use it). Hence, MSG is now roughly 30 times faster and I think
272 that by rewriting a little bit MSG, I could event speed it up a little
273 bit more. Beside the gain in speed, it is also much easier to encode a
274 new platform model with SURF than it was with SG. More to come...
276 * REVOLUTION 3: I have tried to change a little as possible the API of
277 MSG but a few things really had to disappear. The main differences
278 with the previous version are :
279 1) no more m_links_t and the corresponding functions. Platforms are
280 directly read from a XML description and cannot be hard-coded
281 anymore. The same format is used for application deployment
282 description. The new format is described in the documentation.
283 Have a look in tools/platform_generation. There is a tiny script
284 that converts from the old platform format to the new one. Concerning
285 the application deployment format, parsing the old one is tricky.
286 I think most of you should however be able to convert your files. If
287 it is really an issue, I can write a C code that does the conversion.
289 2) the toolbox tbx does not exist anymore. We now have a library
290 with much more data-structures but without the hash-tables (we have
291 dictionaries that are much faster).
293 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
295 *****************************************************************************
296 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
297 * don't feel like converting the oldies. (Mt) *
298 *****************************************************************************
301 Version 2.90: "the long awaited one"
302 - Finished rewriting and debugging MSG. Rewrote the documentation.
303 - disable GRAS for now since it needs to be ported to the newest SG
306 - Finish the port to windows (using mingw32 for cross-compile)
309 - Main loop and datastructures of SURF. A cpu resource object is
310 functional. Surf can thus be used to create cpu's with variable
311 performance on which you can execute some actions.
313 2004-11-15 Martin Quinson
314 - Port to ARM. Simply added the alignment and size descriptions. Should
315 work, but the ARM machines are so slow that I didn't had the opportunity
316 to 'make check' over there yet.
318 2004-11-15 Arnaud Legrand
319 - Trace manager now written. It uses a heap structure and is therefore
320 expected to be efficient. It may however be speeded up (particularly
321 when many events occur at the same date) by using red and black
322 trees. One day maybe...
323 - Max-min linear system solver written. It uses a sparse matrix
324 structure taking advantage of its expected use. Most operations are
325 O(1) and free/calloc are called as few as possible. The computation of
326 the minimum could however be improved by using a red and black tree
329 2004-11-03 Arnaud Legrand
330 - Rename every gras_* function that was in xbt/ to its xbt_
332 - Add a heap and a doubly-linked list to xbt
333 - Added a dichotomy to the dictionaries. make check works as well before
334 so I assume that the patch is correct. I do not know however if things
335 run effectively faster than before now. :)
337 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
338 SimGrid, and the version number is bumped to 2.x
340 2004-10-29 Martin Quinson
341 - Introduction of the remote errors.
342 They are the result of a RMI/RPC on the remote machine.
343 ErrCodes being scalar values, you can't get the host on which those
344 errors did happen. Extending the error mechanism as in Gnome is possible.
345 No idea yet whether it is a good idea.
347 2004-10-28 Martin Quinson
348 - Interface revolution: the Starred Structure Eradication.
349 I used to do typedef struct {} toto_t; and then handle *toto_t.
350 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
351 - ???_t is a valid type (builded with typedef)
352 - s_toto_t is a structure (access to fields with .)
353 - s_toto is a structure needing 'struct' keyword to be used
354 - e_toto_t is an enum
355 - toto_t is an 'object' (struct*)
357 typedef struct s_toto {} s_toto_t, *toto_t;
358 typedef enum {} e_toto_t;
359 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
360 s_toto_t) is private.
362 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
363 so that it can set the variable to NULL. It was so for dicts and sets,
364 it changed for dynars.
366 - Fix a bunch of memleaks in dict_remove
367 - Fix a bug in sg/server_socket opening: it failed all the time.
369 2004-10-07 Martin Quinson
370 - Speed up dynar lookup operation a bit.
372 gras_dynar_get is dead.
374 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
375 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
376 (which gives you the address of the stored data).
378 gras_dynar_get_as is an helpful macro which allows you to retrieve a
379 copy of the data using an affectation to do the job and not a memcpy.
381 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
382 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
384 It does not really speedup the dynar test because they are
385 setting elements all the time (and look them seldom). But the dict does
386 far more lookup than setting.
388 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
390 2004-10-05 Martin Quinson
391 - Allow to (en/dis)able the cycle detection at run time.
393 Whether we should check for cycle or not is now a property of each
394 datatype. When you think there may be some cycle, use datadesc_cycle_set.
395 datadesc_cycle_unset allow to remove this property when previously set.
397 Note that the cycle detection is off by default since it impacts the
398 performance. Watch the data you feed GRAS with ;)
400 This property is hereditary. Any element embedded in a structure having it
401 set have it set for the time of this data exchange.
403 You should set it both on sender and receiver side. If you don't set it on
404 sender side, it will enter an endless loop. If you forget on receiver
405 side, the cycles won't be recreated after communication.
407 - Header reorganization.
408 Kill gras_private.h, each submodule must load the headers it needs.
410 2004-10-04 Martin Quinson
411 - Interface revolution: do not try to survive to malloc failure.
413 Now, gras_malloc and friends call gras_abort() on failure.
414 As a conclusion, malloc_error is not a valid error anymore, and all
415 functions for which it was the only gras_error_t return value are
416 changed. They now return void, or there result directly.
417 This simplify the API a lot.
419 2004-09-29 Martin Quinson
420 - Re-enable raw sockets.
421 Created by gras_socket_{client,server}_ext;
422 Used with gras_raw_{send,recv}
425 It should allow to kill the last bits of gras first version soon.
427 This is not completely satisfactory yet (duplicate code with
428 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
431 - Simplify transport plugin (internal) interface by not passing any
432 argument to _server and _client, but embedding them in the socket
435 2004-09-28 Martin Quinson
436 - Finish the port to AIX.
437 autoconf was my problem (segfault within the malloc replacement
438 function. No idea why)
440 2004-09-16 Martin Quinson
441 - Fix some size_t madness on 64bit architectures.
443 2004-09-08 Martin Quinson
444 - Reduce the number of system headers loaded, overload some more system
445 calls (such as malloc to cast the result of the system one, and work
447 - Fix and reintroduce the config support
449 2004-09-07 Martin Quinson
450 - Source code reorganization to allow Arnaud to surf all over there.
451 - Allow to document the logging categories.
452 - Remove all uppercase from logging categories and useless cleanup in names.
454 2004-08-18 Martin Quinson
455 Version 0.6.2 (protocol not changed; API changed)
456 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
457 gras_error_t), and NULL when not found. Functions expecting a msgtype
458 as argument (msg_wait; msg_send) deal with NULL argument by providing a
459 hopefully usefull message.
460 - Portability to prehistoric sparcs again
462 2004-08-17 Martin Quinson
463 Version 0.6.1 (protocol not changed; ABI not changed)
464 - prealloc some buffers to speed things up
466 2004-08-11 Martin Quinson
467 Version 0.6 (protocol not changed; ABI expended)
468 - The parsing macro can deal with the references, provided that you add
469 the relevant annotations (using GRAS_ANNOTE(size,field_name))
471 2004-08-09 Martin Quinson
472 Version 0.5 (protocol not changed; ABI changed)
473 - Allow to off turn the cycle detection code in data exchange at
474 compilation time. It should be at run time, but I'm short of time (and
475 the config stuff is still broken). That way, we keep dict out of the
476 critical path, which is good because the performance is poor:
477 - search not dichotomial yet
478 - dynar give no way to access their content and memcpy everytime
479 - In composed data description (struct, ref and so on), stop foolness of
480 keeping the subtype's ID, but store the type itself. This keeps sets out
481 of the critical path, which is good since they rely on dynar and
482 dictionnaries. The only loose of that is that we cannot detect the
483 redeclaration of a structure/union with another content (but I'm not sure
484 the code detected well this error before anyway). We still can detect
485 the redefinition discrepancy for the other types.
486 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
487 it breaks the code because of type-punning used all over the place).
488 This breaks on all non-gcc architectures (for now).
490 All those changes (plus the buffer of last time) allow me to gain 2 order
491 of magnitude on cruel tests consisting of 800000 array of integers on two
492 level of a hierarchical structure (200 secondes -> 4 secondes)
495 - the selector of reference must now return the type it points to, not
498 2004-08-06 Martin Quinson
499 Version 0.4 (protocol changed; ABI not changed)
500 - Allow to pass --gras-log argument to processes in simulation mode. Really.
501 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
502 - Implement a buffer transport, and use it by default (it relies on tcp in
503 real life and on sg in simulation).
504 That's a bit hackish since I had a new field to the structure to store
505 its data without interfering with the subtype ones. Inheritance
506 is tricky in C. And that's a kind of reverse inheritance with one class
507 derivating two classes. Or maybe a game with java interfaces. Anyway,
508 that's damn hard in C (at least).
509 Moreover, I got tired while trying to ensure plugin separation and
510 genericity in SG mode. MSG wants me to do weird things, so let's go for
511 cruel hacks (temporarily of course ;).
512 See comment in transport_private.h:71
513 - do not include all the _interface headers in private but in the files
514 which really need them (to cut the compilation time when they are
517 2004-07-26 Martin Quinson
518 Version 0.3 (protocol not changed; ABI changed)
519 - Major overhault of the datadesc interface to simplify it:
520 - shorted the function names:
521 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
522 - add a trivial way to push/pop integers into the cbps without malloc.
523 This allows to make really generic sub_type description, which simply
524 pop their size of the stack.
525 - add a function gras_datadesc_ref_pop_arr() which does what users want
526 most of the time: Declare a dynamic array (which pops its size of the
527 stack) and declare a reference to it. Poor name, but anyway.
528 - kill the post-send callback, add a post-receive one
530 2004-07-23 Martin Quinson
531 Version 0.2 (protocol changed; ABI changed)
532 - add some testing for cpbs in the test cases, and fix some more bugs.
533 This invalidate again the little64 data file, since I cannot regenerate
535 - remove an awfull optimization in the logging stuff, allowing me to:
536 - understand it again
537 - learn gcc how to check that the argument match the provided format
538 - fix all errors revealed by gcc after that
539 - internal keys of dict are not \0 terminated. Deal with it properly in
540 loggings instead of segfaulting when the user want to see the logs :-/
542 2004-07-22 Martin Quinson
543 - Fix some stupid bug preventing cbps (callback postit) from working
545 2004-07-21 Martin Quinson
546 - Some documentation cleanups
547 - remove the useless last argument of msgtype_declare
548 - rename the Virtu functions to fit into the 'os' namespace
549 - move headers src/include -> src/include/gras/ and stop fooling with
550 gras -> . symbolic link
551 - make distcheck is now successful
553 2004-07-19 Martin Quinson
555 - Build shared library also
556 - Install html doc to the right location
557 - stop removing maintainer files in make clean
558 - build tests only on make check
560 2004-07-13 Martin Quinson
562 - No major issue in previous version => change versionning schema
563 - Re-enable little64 convertion test now that Abdou kindly regenerated the
564 corresponding dataset.
566 2004-07-11 Martin Quinson
568 - Get it working with any kind of structure (we can compute the padding
569 bytes remotely for all the architectures I have access to)
570 - Implement the structure parsing macro (still not quite robust/complete)
571 - Improvement to the remote testing toysuite
573 2004-07-10 Martin Quinson
575 - get ride of a bunch of deprecated macros
576 - actually run the test for two-compliment, not only compile it :-/
577 - test whether the structures get packed (and bail out if yes. Damn.
578 Alignment is a serious matter)
579 - test whether the structures get compacted (but respecting the alignment
580 constraints of each types)
581 - test whether the array fields of structures can straddle alignment boundaries
583 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
584 (took me 2 hours to find that bug, looking at the wrong place)
585 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
586 before sending/receiving and are used to compute the offsets of fields
587 - Given that padding size depend even on compiler options, keep track of
588 alignment and aligned_size only for the current architecture. Not a big
589 deal since we send structure fields one after the other (seems
591 - Add the datastructure used for IEEE paper by the PBIO guys to the test
592 program, let it work on linux/gcc/little32. portability todo.
594 2004-07-08 Martin Quinson
595 - import and improve remote compilation support from FAST
596 - make sure make check works on half a dozen of machines out there
598 2004-07-07 Martin Quinson
599 Let's say it's version 0.0.3 ;)
600 - Implement conversions (yuhu!)
601 - Let it work on solaris (beside conversion, of course)
602 - Stupid me, using rand() to generate the conversion datatests in not wise.
604 2004-07-06 Martin Quinson
605 - Let make dist work, since I'm gonna need it to compile on remote hosts
606 - Let Tests/datadesc_usage write the architecture on which the file was
607 generated as first byte.
608 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
609 architecture descriptions.
610 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
611 on those architectures.
612 - Optimization: send/recv array of scalar in one shoot
614 2004-07-05 Martin Quinson
615 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
617 - Plug a whole bunch of memleaks
618 - each process now have to call gras_{init,exit}. One day, their log
619 settings will be separated
620 - Continue the code factorisation between SG, RL and common in Transport.
622 2004-07-04 Martin Quinson
624 - Redistribution between SG and RL.
625 We wanna have to accept in SG, so move accepted related parts of RL in
626 the common part. (more precisely, the dynar of all known sockets is no
627 more a static in transport.c, but part of the process_data)
629 [gras_stub_generator]
630 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
632 2004-07-03 Martin Quinson
633 - Create a new log channel tbx containing dict, set, log, dynar (to shut
634 them all up in one shot)
636 - Fix the ugly case of reference to dynamic array.
637 - New (semi-public) function gras_datadesc_size to allow the messaging
638 layer to malloc the needed space for the buffer.
640 - gras_socket_close now expect the socket to close (and not its address to
641 put NULL in it after it). This is because the socket passed to handlers
642 is one of their argument (=> not writable).
644 - propagate the interface cleanup from last week in datadesc, ie remove a
645 superfluous level of indirection. User pass adress of variable
646 containing data (both when sending and receiving), and not of a variable
647 being a pointer to the data. Let's say that I like it better ;)
648 The price for that is constructs like "int msg=*(int*)payload" in
649 handlers, but it's a fine price, IMHO.
651 - Let it work in RL (yuhu)
653 2004-06-21 Martin Quinson
655 - porting SG plugin and SG select to new standards (works almost).
656 - plug memleaks and fix bugs around.
659 - cleanup the prototype of data recv and force users to specify when they
660 want to handle references to objects. Test case working even for cycles.
661 - plug memleaks. Valgrind is perfectly ok with this.
663 2004-06-12 Martin Quinson
665 - cleanup the separation between plugin and main code in plugin creation
667 2004-06-11 Martin Quinson
669 - Reput hook for raw sockets, needed for BW experiments
670 - kill a few lines of dead code
671 [Data description] Interface cleanup
672 - gras_datadesc_by_name returns the searched type or NULL.
673 That way, no variable is needed to use a type desc once, which makes
675 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
676 two parameters were strings (field name, type name), leading to
678 [Dicos] Interface cleanup
679 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
680 This is consistant with the dynar API.
682 2004-04-21 Martin Quinson
684 - Porting to new standards.
687 There is no bag anymore, no need to take extra provision to mask the
688 pointers behind "ID".
689 Better splitup of functions between files create/exchange/convert.
690 This is still a bit artificial since convert and receive are so
691 interleaved, but anyway.
693 - add a queued message list to procdata (the ones not matching criteria
695 - factorize some more code between SG and RL wrt procdata
697 - use gras_exit in example to track memleaks
698 - get rid of gs_example now that GS is properly integrated into gras
699 - update run_test to integrate the lastest tests (datadesc)
701 - rename WARNINGn macros to WARNn since it prooved error-prone
703 2004-04-19 Martin Quinson
705 - register init/exit functions within gras module mechanism
706 - send/receive function.
707 Convertion is not implemented, but short-cutted if not needed.
708 struct/array elements are sent one by one (instead of block-wise), but
709 nobody really cares (yet). Get a prototype before optimizing.
710 - tests (using a file socket) for DD send/receive on:
711 - base types: int, float
712 - array: fixed size, string (ie ref to dynamic string)
713 - structure: homogeneous, heterogeneous
714 - chained list, graph with cycle
715 Believe it or not, valgrind is not too unhappy with the results. The
716 cycle happily segfaults, but the others are ok. And I'm sick of pointers
720 - Bugfix when using a filename explicitely (instead of '-')
722 2004-04-09 Martin Quinson
724 - factorize more code between RL and SG in socket creation
725 - Complete the implementation and tests of:
727 o file (only in RL, and mainly for debugging)
729 I lost 3 days to design a portable address resolver, and then decided
730 that the prototype mainly have to run on my box.
731 Addressing portability too early may be like optimizing too early :-/
733 - use gras_init in the Tests instead of the crappy parse_log_opt
734 (the latter function is removed)
735 [Conditional execution]
736 - New functions: gras_if_RL/gras_if_SG (basic support for this)
737 [Code reorganisation]
738 - Get rid of libgrasutils.a since it makes more trouble than it solves.
739 Build examples against the RL library, since there is no way to disable
740 its creation for now.