1 SimGrid (2.96) unstable; urgency=low
7 SimGrid (2.95) unstable; urgency=low
10 * Steal some nice code to GNU pth to fix context detection and usage [AL]
11 * Cleanup in the xbt_config API; add configuration callbacks. [MQ]
12 * Cleanup in the initialization API: the unused "defaultlog" is dead. [MQ]
15 * Bugfix: Allow absolute paths for platform description files [MQ]
16 * Bugfix: do free the variables after use. Leads to drastic performance
18 * Implement max_duration (ie, timeouts) on resources [AL]
21 * Implement MSG_config to configure MSG at runtime. xbt_cfg test on a real
23 * Implement MSG_channel_select_from() to help GRAS now that SURF provide
24 the needed support (timeouts) [AL]
27 * Implement measurement sockets. You can now get the bandwidth between two
28 hosts thanks to AMOK (see below). [MQ]
29 * gras_datadesc_dynar() builds a dynar type descriptor, allowing to send
30 dynar over the network (yeah) [MQ]
31 * Real (even if simplistic) implementation of gras_os_myname() on RL [MQ]
32 * simple/static token-ring example. [Alexandre Colucci and MQ]
33 * Use MSG_channel_select_from() and remove the *slow* hack we had to put
37 * Differentiate the types "char[22]" and "unsigned char[22]" in automatic
38 type parsing. "short" and "long" modifiers were also ignored; other
39 modifier (such as reference level) are still ignored. [MQ]
40 * Embeed the buffer size within the buffer itself on SG. [MQ]
41 That way, send() are atomic and cannot get intermixed anymore (at least
42 the ones which are less than 100k; bigger messages still have the issue)
43 * Array size pushed by the field, not by the field type (or each
44 and every long int will push stuff to the cbps) [MQ]
45 * use select() to sleep since it allows to portably sleep less than one
49 * <project>.Makefile.local (generated from gras_stub_generator) |MQ]:
52 * Type Callbacks now receive the gras_datadesc_type_t they work on as argument.
53 * type category 'ignored' killed as it was never used and were difficult
55 * whether a type can cycle or not is now a flag, leaving room for more
56 flags (as "ignored", if we feel the need one day ;)
57 * Rename raw sockets to measurement sockets since "raw" has another
58 meaning in networking community.
61 * Advanced Metacomputing Overlay Kit introduction. It is based over GRAS
62 and offers features not belonging to GRAS but that most applications
63 need. One day, it may be a set of plugins loadable at runtime.
64 * New module: bandwidth
65 bandwidth measurement between arbitrary nodes running this module. [MQ]
67 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Thu, 30 Jun 2005 16:29:20 -0700
69 SimGrid (2.94) unstable; urgency=low
71 The first beta release of SimGrid 3 !
75 * Update the main page and the FAQ. Adding references to gforge.
78 * Add a gras_os_getpid function.
81 * Add MSG_task_get_compute_duration() and MSG_task_get_data_size()
82 * Extend the logs so that they also print PID, hostname, date, ... if
84 * Convert the MSG example to the use of xbt_logs instead of PRINT_MESSAGE,
85 and kill the old version which were in testsuite/
86 * Rewrite tools/MSG_visualization/colorize.pl for using with logs instead
90 * Add xbt_os_time(). As the rest of xbt/portability, this is not public
91 for users. Instead, each programming environment (GRAS, MSG,...) use it
92 when needed to provide such a feature to users.
93 Don't shortcut the mecanism or you will also shortcut the virtualization
94 you need on the simulator.
98 * Cleanups in configury with regard to compile optimization/warning flags.
99 Also add -fno-loop-optimize to any powerpc since it's the optimization
100 killing gcc (< 3.4.0).
101 * Doxygen cleanups: move MSG examples, kill the second Doxygen phase
102 needed by MSG examples complications
103 * Borrow configury beautifications from PHP
106 * Bugfix: XBT_LOG_NEW_DEFAULT_CATEGORY now compiles without compiler
107 warning (thanks loris for stumbling into this one).
108 * Bugfix: stop loading private headers (gras_config.h) from the public
112 * Change SIMGRID_INSTALL_PATH to GRAS_ROOT in Makefiles generated for user.
113 * Rename gras_get_my_fqdn to gras_os_myname and implement it in the simulator
114 RL would imply a DNS resolver, which is *hard* to do in a portable way
115 (and therefore delayed).
116 * Implement a real timer mecanism and use it in timing macros. This allows
117 to avoid rounding errors and get a 0.000005 sec precision in timing
118 macros. While I was at it, various cleanups:
119 - allow to declare more than one timed section per file (fix a stupid bug)
120 - move some private declaration to the right place
121 - merge conditional execution and timing macros into emulation module
122 - document the module
123 - make sure the module cleanups its mess on gras_exit
124 * Documentation improvements:
125 - (new) how to compile applications using GRAS
126 - (new) emulation support (timing macros)
128 -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr> Fri, 13 May 2005 10:49:31 +0200
130 SimGrid (2.93) unstable; urgency=low
132 Alpha 4 on the path to SimGrid 3 (aka the "neuf-trois" version)
135 - Use Paje properly where used. Still to be sanitized properly.
136 - Portability fix: Add an implementation of the contexts using pthread
140 - Add xbt_procname(): returns the name of the current process.
141 Use it to show the current process's name in all logging.
143 - fix detection of older flex version and the reaction, since we do
144 depend on modern ones (we use lex_destroy)
145 - Better separation of SG and RL in the libs: remove all simulation code
146 from libgras. As a result, this lib is now only 200k when stripped.
147 Some of the xbt modules may also be duplicated (two sets and such) and
148 should be cleaned/killed before SG3.
149 - Insist on using xlC on AIX because of weird problems involving gcc there.
150 - Cleanup the make remote stuff. This is now done by scripts
151 tools/graspe-{master,slave} (GRAS Platform Expender). This is still
152 mainly for our private use, but we're working on changing them to user
155 - Bugfix: flush the socket on close only if there is some *output*.
156 - Bugfix: flush idempotent when there's nothing to send (don't send size=0)
158 - Add MSG_task_get_name. The task names are mainly for debugging purpose,
161 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Fri, 4 Mar 2005 14:32:37 -0800
163 SimGrid (2.92) unstable; urgency=low
165 Alpha 3 on the path to SimGrid 3
169 - New! First try of benchmarking macros.
170 - New! First try so that gras_stub_generator generate deployment and
171 remote compilation helpers.
173 - Bugfix: Initialization fix in msg_test.
177 - Bugfix: applied patch to lexer so that it doesn't need a huge heap.
179 - Bugfix: let dicts work with NULL content (_foreach didn't) and cleanups
181 - API Change: gras_os_sleep to take the amount of seconds as a double.
182 Accepting an int was error prone since it was the only location where
183 seconds were coded as such. It leaded to damn rounding errors.
184 - Bugfix: Hard to belive that timers ever worked before this.
186 -- SimGrid team <simgrid2-users@listes.ens-lyon.fr> Wed, 23 Feb 2005 22:09:21 +0100
188 SimGrid (2.91) unstable; urgency=low
190 Alpha 2 on the path to SimGrid 3
194 - Bug fix in the lmm_solver.
196 - New! Interface to Paje (see http://www-id.imag.fr/Logiciels/paje/)
197 through the function MSG_paje_output.
198 - New! Introducing two new functions MSG_process_kill() and MSG_process_killall().
199 - It is possible to bound the rate of a communication in MSG with
200 MSG_task_put_bounded() (was already in the previous version but I had forgotten
201 to write it in the changelog).
202 - Bug fix to let GRAS run on top of MSG until we move it directly on top
207 - Various cleanups to the autotools stuff
208 - Begin to move Gras examples to examples/gras/
209 - Let make distcheck work again (yeah!)
211 - documentation overhauled using doxygen.
212 gtk-doc-tools is dead in SimGrid now.
213 - Automatically extract all existing logging categories, and add the list
214 to the documentation (long standing one, to say the less)
216 - Cleanup the known architecture table. Reorder the entries to group what
217 should be, and use a more consistent naming scheme.
218 (some of the test dataset are still to be regenerated)
219 - New! Allow library to register globals on each process just as userdata
221 This is implemented using a xbt_dict and not a xbt_set, so we loose the
222 lookup time (for now).
223 Use it in msg and trp.
224 This cleans a lot the internals and helps enforcing privacy of the
225 headers between the gras components.
226 - New! Add a timer mechanism, not unlike cron(8) and at(1).
227 - Bugfix: gras_os_time was delirious in RL.
228 - Bugfix: gras_trp_select/RL don't run into the wall when asked to select
230 - Reenable GRAS now that it works.
232 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 14 Feb 2005 14:02:13 -0800
234 SimGrid (2.90) unstable; urgency=low
236 Alpha 1 on the path to SimGrid 3
238 * It is a long time since the last release of SimGrid. I'm sorry about
239 that but as I had told you, I was rewriting a lot of things. I apologize
240 to those who had been reporting bugs to me and that I had not answered.
241 If your bug is still in the new version, please tell me. Here is a
242 summary of the main changes.
244 * REVOLUTION 1: The SimGrid project has merged with the GRAS project
245 lead by Martin Quinson. As a consequence SimGrid gains a lot in
246 portability, speed, and a lot more but you'll figure it out later.
247 SimGrid now comprises 3 different projects : MSG, GRAS and SMPI.
248 I wanted to release the new MSG as soon as possible and I have
249 broken GRAS, which is the reason why, for now, only MSG is fully
250 functional. A laconic description of these projects is available
251 in the documentation.
253 * REVOLUTION 2: I have removed SG and I am now using a new simulation
254 kernel optimized for our needs (called SURF but only the developers
255 should use it). Hence, MSG is now roughly 30 times faster and I think
256 that by rewriting a little bit MSG, I could event speed it up a little
257 bit more. Beside the gain in speed, it is also much easier to encode a
258 new platform model with SURF than it was with SG. More to come...
260 * REVOLUTION 3: I have tried to change a little as possible the API of
261 MSG but a few things really had to disappear. The main differences
262 with the previous version are :
263 1) no more m_links_t and the corresponding functions. Platforms are
264 directly read from a XML description and cannot be hard-coded
265 anymore. The same format is used for application deployment
266 description. The new format is described in the documentation.
267 Have a look in tools/platform_generation. There is a tiny script
268 that converts from the old platform format to the new one. Concerning
269 the application deployment format, parsing the old one is tricky.
270 I think most of you should however be able to convert your files. If
271 it is really an issue, I can write a C code that does the conversion.
273 2) the toolbox tbx does not exist anymore. We now have a library
274 with much more data-structures but without the hash-tables (we have
275 dictionaries that are much faster).
277 -- Arnaud Legrand <Arnaud.Legrand@imag.fr> Mon, 31 Jan 2005 10:45:53 -0800
279 *****************************************************************************
280 * Follows the old GRAS changelog. It does not follow the same syntax, but I *
281 * don't feel like converting the oldies. (Mt) *
282 *****************************************************************************
285 Version 2.90: "the long awaited one"
286 - Finished rewriting and debugging MSG. Rewrote the documentation.
287 - disable GRAS for now since it needs to be ported to the newest SG
290 - Finish the port to windows (using mingw32 for cross-compile)
293 - Main loop and datastructures of SURF. A cpu resource object is
294 functional. Surf can thus be used to create cpu's with variable
295 performance on which you can execute some actions.
297 2004-11-15 Martin Quinson
298 - Port to ARM. Simply added the alignment and size descriptions. Should
299 work, but the ARM machines are so slow that I didn't had the opportunity
300 to 'make check' over there yet.
302 2004-11-15 Arnaud Legrand
303 - Trace manager now written. It uses a heap structure and is therefore
304 expected to be efficient. It may however be speeded up (particularly
305 when many events occur at the same date) by using red and black
306 trees. One day maybe...
307 - Max-min linear system solver written. It uses a sparse matrix
308 structure taking advantage of its expected use. Most operations are
309 O(1) and free/calloc are called as few as possible. The computation of
310 the minimum could however be improved by using a red and black tree
313 2004-11-03 Arnaud Legrand
314 - Rename every gras_* function that was in xbt/ to its xbt_
316 - Add a heap and a doubly-linked list to xbt
317 - Added a dichotomy to the dictionaries. make check works as well before
318 so I assume that the patch is correct. I do not know however if things
319 run effectively faster than before now. :)
321 Inclusion of the SimGrid tree in the GRAS one. The archive is renamed to
322 SimGrid, and the version number is bumped to 2.x
324 2004-10-29 Martin Quinson
325 - Introduction of the remote errors.
326 They are the result of a RMI/RPC on the remote machine.
327 ErrCodes being scalar values, you can't get the host on which those
328 errors did happen. Extending the error mechanism as in Gnome is possible.
329 No idea yet whether it is a good idea.
331 2004-10-28 Martin Quinson
332 - Interface revolution: the Starred Structure Eradication.
333 I used to do typedef struct {} toto_t; and then handle *toto_t.
334 Arnaud (and Oli) didn't like it, and I surrendered. Now, you have:
335 - ???_t is a valid type (builded with typedef)
336 - s_toto_t is a structure (access to fields with .)
337 - s_toto is a structure needing 'struct' keyword to be used
338 - e_toto_t is an enum
339 - toto_t is an 'object' (struct*)
341 typedef struct s_toto {} s_toto_t, *toto_t;
342 typedef enum {} e_toto_t;
343 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly
344 s_toto_t) is private.
346 - While I was at it, all gras_<obj>_free() functions want a gras_<obj>_t*
347 so that it can set the variable to NULL. It was so for dicts and sets,
348 it changed for dynars.
350 - Fix a bunch of memleaks in dict_remove
351 - Fix a bug in sg/server_socket opening: it failed all the time.
353 2004-10-07 Martin Quinson
354 - Speed up dynar lookup operation a bit.
356 gras_dynar_get is dead.
358 Now, you can choose between gras_dynar_get_cpy (the old gras_dynar_get
359 but should be avoided for efficiency reasons) and gras_dynar_get_ptr
360 (which gives you the address of the stored data).
362 gras_dynar_get_as is an helpful macro which allows you to retrieve a
363 copy of the data using an affectation to do the job and not a memcpy.
365 int toto = gras_dynar_get_as(dyn,0,int); rewrites itself to
366 int toto = *(int*)gras_dynar_get_ptr(dyn,0);
368 It does not really speedup the dynar test because they are
369 setting elements all the time (and look them seldom). But the dict does
370 far more lookup than setting.
372 So, this brings the dict_crash test from ~33s to ~25s (200000 elms).
374 2004-10-05 Martin Quinson
375 - Allow to (en/dis)able the cycle detection at run time.
377 Whether we should check for cycle or not is now a property of each
378 datatype. When you think there may be some cycle, use datadesc_cycle_set.
379 datadesc_cycle_unset allow to remove this property when previously set.
381 Note that the cycle detection is off by default since it impacts the
382 performance. Watch the data you feed GRAS with ;)
384 This property is hereditary. Any element embedded in a structure having it
385 set have it set for the time of this data exchange.
387 You should set it both on sender and receiver side. If you don't set it on
388 sender side, it will enter an endless loop. If you forget on receiver
389 side, the cycles won't be recreated after communication.
391 - Header reorganization.
392 Kill gras_private.h, each submodule must load the headers it needs.
394 2004-10-04 Martin Quinson
395 - Interface revolution: do not try to survive to malloc failure.
397 Now, gras_malloc and friends call gras_abort() on failure.
398 As a conclusion, malloc_error is not a valid error anymore, and all
399 functions for which it was the only gras_error_t return value are
400 changed. They now return void, or there result directly.
401 This simplify the API a lot.
403 2004-09-29 Martin Quinson
404 - Re-enable raw sockets.
405 Created by gras_socket_{client,server}_ext;
406 Used with gras_raw_{send,recv}
409 It should allow to kill the last bits of gras first version soon.
411 This is not completely satisfactory yet (duplicate code with
412 chunk_{send,recv}; a bit out of the plugin mechanism), but it should
415 - Simplify transport plugin (internal) interface by not passing any
416 argument to _server and _client, but embedding them in the socket
419 2004-09-28 Martin Quinson
420 - Finish the port to AIX.
421 autoconf was my problem (segfault within the malloc replacement
422 function. No idea why)
424 2004-09-16 Martin Quinson
425 - Fix some size_t madness on 64bit architectures.
427 2004-09-08 Martin Quinson
428 - Reduce the number of system headers loaded, overload some more system
429 calls (such as malloc to cast the result of the system one, and work
431 - Fix and reintroduce the config support
433 2004-09-07 Martin Quinson
434 - Source code reorganization to allow Arnaud to surf all over there.
435 - Allow to document the logging categories.
436 - Remove all uppercase from logging categories and useless cleanup in names.
438 2004-08-18 Martin Quinson
439 Version 0.6.2 (protocol not changed; API changed)
440 - Interface cleanup: gras_msgtype_by_name returns the type (instead of a
441 gras_error_t), and NULL when not found. Functions expecting a msgtype
442 as argument (msg_wait; msg_send) deal with NULL argument by providing a
443 hopefully usefull message.
444 - Portability to prehistoric sparcs again
446 2004-08-17 Martin Quinson
447 Version 0.6.1 (protocol not changed; ABI not changed)
448 - prealloc some buffers to speed things up
450 2004-08-11 Martin Quinson
451 Version 0.6 (protocol not changed; ABI expended)
452 - The parsing macro can deal with the references, provided that you add
453 the relevant annotations (using GRAS_ANNOTE(size,field_name))
455 2004-08-09 Martin Quinson
456 Version 0.5 (protocol not changed; ABI changed)
457 - Allow to off turn the cycle detection code in data exchange at
458 compilation time. It should be at run time, but I'm short of time (and
459 the config stuff is still broken). That way, we keep dict out of the
460 critical path, which is good because the performance is poor:
461 - search not dichotomial yet
462 - dynar give no way to access their content and memcpy everytime
463 - In composed data description (struct, ref and so on), stop foolness of
464 keeping the subtype's ID, but store the type itself. This keeps sets out
465 of the critical path, which is good since they rely on dynar and
466 dictionnaries. The only loose of that is that we cannot detect the
467 redeclaration of a structure/union with another content (but I'm not sure
468 the code detected well this error before anyway). We still can detect
469 the redefinition discrepancy for the other types.
470 - Use a whole bunch of optimisation flags (plus -fno-strict-aliasing since
471 it breaks the code because of type-punning used all over the place).
472 This breaks on all non-gcc architectures (for now).
474 All those changes (plus the buffer of last time) allow me to gain 2 order
475 of magnitude on cruel tests consisting of 800000 array of integers on two
476 level of a hierarchical structure (200 secondes -> 4 secondes)
479 - the selector of reference must now return the type it points to, not
482 2004-08-06 Martin Quinson
483 Version 0.4 (protocol changed; ABI not changed)
484 - Allow to pass --gras-log argument to processes in simulation mode. Really.
485 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
486 - Implement a buffer transport, and use it by default (it relies on tcp in
487 real life and on sg in simulation).
488 That's a bit hackish since I had a new field to the structure to store
489 its data without interfering with the subtype ones. Inheritance
490 is tricky in C. And that's a kind of reverse inheritance with one class
491 derivating two classes. Or maybe a game with java interfaces. Anyway,
492 that's damn hard in C (at least).
493 Moreover, I got tired while trying to ensure plugin separation and
494 genericity in SG mode. MSG wants me to do weird things, so let's go for
495 cruel hacks (temporarily of course ;).
496 See comment in transport_private.h:71
497 - do not include all the _interface headers in private but in the files
498 which really need them (to cut the compilation time when they are
501 2004-07-26 Martin Quinson
502 Version 0.3 (protocol not changed; ABI changed)
503 - Major overhault of the datadesc interface to simplify it:
504 - shorted the function names:
505 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
506 - add a trivial way to push/pop integers into the cbps without malloc.
507 This allows to make really generic sub_type description, which simply
508 pop their size of the stack.
509 - add a function gras_datadesc_ref_pop_arr() which does what users want
510 most of the time: Declare a dynamic array (which pops its size of the
511 stack) and declare a reference to it. Poor name, but anyway.
512 - kill the post-send callback, add a post-receive one
514 2004-07-23 Martin Quinson
515 Version 0.2 (protocol changed; ABI changed)
516 - add some testing for cpbs in the test cases, and fix some more bugs.
517 This invalidate again the little64 data file, since I cannot regenerate
519 - remove an awfull optimization in the logging stuff, allowing me to:
520 - understand it again
521 - learn gcc how to check that the argument match the provided format
522 - fix all errors revealed by gcc after that
523 - internal keys of dict are not \0 terminated. Deal with it properly in
524 loggings instead of segfaulting when the user want to see the logs :-/
526 2004-07-22 Martin Quinson
527 - Fix some stupid bug preventing cbps (callback postit) from working
529 2004-07-21 Martin Quinson
530 - Some documentation cleanups
531 - remove the useless last argument of msgtype_declare
532 - rename the Virtu functions to fit into the 'os' namespace
533 - move headers src/include -> src/include/gras/ and stop fooling with
534 gras -> . symbolic link
535 - make distcheck is now successful
537 2004-07-19 Martin Quinson
539 - Build shared library also
540 - Install html doc to the right location
541 - stop removing maintainer files in make clean
542 - build tests only on make check
544 2004-07-13 Martin Quinson
546 - No major issue in previous version => change versionning schema
547 - Re-enable little64 convertion test now that Abdou kindly regenerated the
548 corresponding dataset.
550 2004-07-11 Martin Quinson
552 - Get it working with any kind of structure (we can compute the padding
553 bytes remotely for all the architectures I have access to)
554 - Implement the structure parsing macro (still not quite robust/complete)
555 - Improvement to the remote testing toysuite
557 2004-07-10 Martin Quinson
559 - get ride of a bunch of deprecated macros
560 - actually run the test for two-compliment, not only compile it :-/
561 - test whether the structures get packed (and bail out if yes. Damn.
562 Alignment is a serious matter)
563 - test whether the structures get compacted (but respecting the alignment
564 constraints of each types)
565 - test whether the array fields of structures can straddle alignment boundaries
567 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
568 (took me 2 hours to find that bug, looking at the wrong place)
569 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
570 before sending/receiving and are used to compute the offsets of fields
571 - Given that padding size depend even on compiler options, keep track of
572 alignment and aligned_size only for the current architecture. Not a big
573 deal since we send structure fields one after the other (seems
575 - Add the datastructure used for IEEE paper by the PBIO guys to the test
576 program, let it work on linux/gcc/little32. portability todo.
578 2004-07-08 Martin Quinson
579 - import and improve remote compilation support from FAST
580 - make sure make check works on half a dozen of machines out there
582 2004-07-07 Martin Quinson
583 Let's say it's version 0.0.3 ;)
584 - Implement conversions (yuhu!)
585 - Let it work on solaris (beside conversion, of course)
586 - Stupid me, using rand() to generate the conversion datatests in not wise.
588 2004-07-06 Martin Quinson
589 - Let make dist work, since I'm gonna need it to compile on remote hosts
590 - Let Tests/datadesc_usage write the architecture on which the file was
591 generated as first byte.
592 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
593 architecture descriptions.
594 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
595 on those architectures.
596 - Optimization: send/recv array of scalar in one shoot
598 2004-07-05 Martin Quinson
599 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
601 - Plug a whole bunch of memleaks
602 - each process now have to call gras_{init,exit}. One day, their log
603 settings will be separated
604 - Continue the code factorisation between SG, RL and common in Transport.
606 2004-07-04 Martin Quinson
608 - Redistribution between SG and RL.
609 We wanna have to accept in SG, so move accepted related parts of RL in
610 the common part. (more precisely, the dynar of all known sockets is no
611 more a static in transport.c, but part of the process_data)
613 [gras_stub_generator]
614 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
616 2004-07-03 Martin Quinson
617 - Create a new log channel tbx containing dict, set, log, dynar (to shut
618 them all up in one shot)
620 - Fix the ugly case of reference to dynamic array.
621 - New (semi-public) function gras_datadesc_size to allow the messaging
622 layer to malloc the needed space for the buffer.
624 - gras_socket_close now expect the socket to close (and not its address to
625 put NULL in it after it). This is because the socket passed to handlers
626 is one of their argument (=> not writable).
628 - propagate the interface cleanup from last week in datadesc, ie remove a
629 superfluous level of indirection. User pass adress of variable
630 containing data (both when sending and receiving), and not of a variable
631 being a pointer to the data. Let's say that I like it better ;)
632 The price for that is constructs like "int msg=*(int*)payload" in
633 handlers, but it's a fine price, IMHO.
635 - Let it work in RL (yuhu)
637 2004-06-21 Martin Quinson
639 - porting SG plugin and SG select to new standards (works almost).
640 - plug memleaks and fix bugs around.
643 - cleanup the prototype of data recv and force users to specify when they
644 want to handle references to objects. Test case working even for cycles.
645 - plug memleaks. Valgrind is perfectly ok with this.
647 2004-06-12 Martin Quinson
649 - cleanup the separation between plugin and main code in plugin creation
651 2004-06-11 Martin Quinson
653 - Reput hook for raw sockets, needed for BW experiments
654 - kill a few lines of dead code
655 [Data description] Interface cleanup
656 - gras_datadesc_by_name returns the searched type or NULL.
657 That way, no variable is needed to use a type desc once, which makes
659 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
660 two parameters were strings (field name, type name), leading to
662 [Dicos] Interface cleanup
663 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
664 This is consistant with the dynar API.
666 2004-04-21 Martin Quinson
668 - Porting to new standards.
671 There is no bag anymore, no need to take extra provision to mask the
672 pointers behind "ID".
673 Better splitup of functions between files create/exchange/convert.
674 This is still a bit artificial since convert and receive are so
675 interleaved, but anyway.
677 - add a queued message list to procdata (the ones not matching criteria
679 - factorize some more code between SG and RL wrt procdata
681 - use gras_exit in example to track memleaks
682 - get rid of gs_example now that GS is properly integrated into gras
683 - update run_test to integrate the lastest tests (datadesc)
685 - rename WARNINGn macros to WARNn since it prooved error-prone
687 2004-04-19 Martin Quinson
689 - register init/exit functions within gras module mechanism
690 - send/receive function.
691 Convertion is not implemented, but short-cutted if not needed.
692 struct/array elements are sent one by one (instead of block-wise), but
693 nobody really cares (yet). Get a prototype before optimizing.
694 - tests (using a file socket) for DD send/receive on:
695 - base types: int, float
696 - array: fixed size, string (ie ref to dynamic string)
697 - structure: homogeneous, heterogeneous
698 - chained list, graph with cycle
699 Believe it or not, valgrind is not too unhappy with the results. The
700 cycle happily segfaults, but the others are ok. And I'm sick of pointers
704 - Bugfix when using a filename explicitely (instead of '-')
706 2004-04-09 Martin Quinson
708 - factorize more code between RL and SG in socket creation
709 - Complete the implementation and tests of:
711 o file (only in RL, and mainly for debugging)
713 I lost 3 days to design a portable address resolver, and then decided
714 that the prototype mainly have to run on my box.
715 Addressing portability too early may be like optimizing too early :-/
717 - use gras_init in the Tests instead of the crappy parse_log_opt
718 (the latter function is removed)
719 [Conditional execution]
720 - New functions: gras_if_RL/gras_if_SG (basic support for this)
721 [Code reorganisation]
722 - Get rid of libgrasutils.a since it makes more trouble than it solves.
723 Build examples against the RL library, since there is no way to disable
724 its creation for now.