1 2004-08-06 Martin Quinson
2 Version 0.4 (protocol changed; ABI not changed)
3 - Allow to pass --gras-log argument to processes in simulation mode. Really.
4 - New debugging level: trace (under debug) to see effect of GRAS_IN/OUT
5 - Implement a buffer transport, and use it by default (it relies on tcp in
6 real life and on sg in simulation).
7 That's a bit hackish since I had a new field to the structure to store
8 its data without interfering with the subtype ones. Inheritance
9 is tricky in C. And that's a kind of reverse inheritance with one class
10 derivating two classes. Or maybe a game with java interfaces. Anyway,
11 that's damn hard in C (at least).
12 Moreover, I got tired while trying to ensure plugin separation and
13 genericity in SG mode. MSG wants me to do weird things, so let's go for
14 cruel hacks (temporarily of course ;).
15 See comment in transport_private.h:71
16 - do not include all the _interface headers in private but in the files
17 which really need them (to cut the compilation time when they are
20 2004-07-26 Martin Quinson
21 Version 0.3 (protocol not changed; ABI changed)
22 - Major overhault of the datadesc interface to simplify it:
23 - shorted the function names:
24 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
25 - add a trivial way to push/pop integers into the cbps without malloc.
26 This allows to make really generic sub_type description, which simply
27 pop their size of the stack.
28 - add a function gras_datadesc_ref_pop_arr() which does what users want
29 most of the time: Declare a dynamic array (which pops its size of the
30 stack) and declare a reference to it. Poor name, but anyway.
31 - kill the post-send callback, add a post-receive one
33 2004-07-23 Martin Quinson
34 Version 0.2 (protocol changed; ABI changed)
35 - add some testing for cpbs in the test cases, and fix some more bugs.
36 This invalidate again the little64 data file, since I cannot regenerate
38 - remove an awfull optimization in the logging stuff, allowing me to:
40 - learn gcc how to check that the argument match the provided format
41 - fix all errors revealed by gcc after that
42 - internal keys of dict are not \0 terminated. Deal with it properly in
43 loggings instead of segfaulting when the user want to see the logs :-/
45 2004-07-22 Martin Quinson
46 - Fix some stupid bug preventing cbps (callback postit) from working
48 2004-07-21 Martin Quinson
49 - Some documentation cleanups
50 - remove the useless last argument of msgtype_declare
51 - rename the Virtu functions to fit into the 'os' namespace
52 - move headers src/include -> src/include/gras/ and stop fooling with
53 gras -> . symbolic link
54 - make distcheck is now successful
56 2004-07-19 Martin Quinson
58 - Build shared library also
59 - Install html doc to the right location
60 - stop removing maintainer files in make clean
61 - build tests only on make check
63 2004-07-13 Martin Quinson
65 - No major issue in previous version => change versionning schema
66 - Re-enable little64 convertion test now that Abdou kindly regenerated the
67 corresponding dataset.
69 2004-07-11 Martin Quinson
71 - Get it working with any kind of structure (we can compute the padding
72 bytes remotely for all the architectures I have access to)
73 - Implement the structure parsing macro (still not quite robust/complete)
74 - Improvement to the remote testing toysuite
76 2004-07-10 Martin Quinson
78 - get ride of a bunch of deprecated macros
79 - actually run the test for two-compliment, not only compile it :-/
80 - test whether the structures get packed (and bail out if yes. Damn.
81 Alignment is a serious matter)
82 - test whether the structures get compacted (but respecting the alignment
83 constraints of each types)
84 - test whether the array fields of structures can straddle alignment boundaries
86 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
87 (took me 2 hours to find that bug, looking at the wrong place)
88 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
89 before sending/receiving and are used to compute the offsets of fields
90 - Given that padding size depend even on compiler options, keep track of
91 alignment and aligned_size only for the current architecture. Not a big
92 deal since we send structure fields one after the other (seems
94 - Add the datastructure used for IEEE paper by the PBIO guys to the test
95 program, let it work on linux/gcc/little32. portability todo.
97 2004-07-08 Martin Quinson
98 - import and improve remote compilation support from FAST
99 - make sure make check works on half a dozen of machines out there
101 2004-07-07 Martin Quinson
102 Let's say it's version 0.0.3 ;)
103 - Implement conversions (yuhu!)
104 - Let it work on solaris (beside conversion, of course)
105 - Stupid me, using rand() to generate the conversion datatests in not wise.
107 2004-07-06 Martin Quinson
108 - Let make dist work, since I'm gonna need it to compile on remote hosts
109 - Let Tests/datadesc_usage write the architecture on which the file was
110 generated as first byte.
111 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
112 architecture descriptions.
113 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
114 on those architectures.
115 - Optimization: send/recv array of scalar in one shoot
117 2004-07-05 Martin Quinson
118 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
120 - Plug a whole bunch of memleaks
121 - each process now have to call gras_{init,exit}. One day, their log
122 settings will be separated
123 - Continue the code factorisation between SG, RL and common in Transport.
125 2004-07-04 Martin Quinson
127 - Redistribution between SG and RL.
128 We wanna have to accept in SG, so move accepted related parts of RL in
129 the common part. (more precisely, the dynar of all known sockets is no
130 more a static in transport.c, but part of the process_data)
132 [gras_stub_generator]
133 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
135 2004-07-03 Martin Quinson
136 - Create a new log channel tbx containing dict, set, log, dynar (to shut
137 them all up in one shot)
139 - Fix the ugly case of reference to dynamic array.
140 - New (semi-public) function gras_datadesc_size to allow the messaging
141 layer to malloc the needed space for the buffer.
143 - gras_socket_close now expect the socket to close (and not its address to
144 put NULL in it after it). This is because the socket passed to handlers
145 is one of their argument (=> not writable).
147 - propagate the interface cleanup from last week in datadesc, ie remove a
148 superfluous level of indirection. User pass adress of variable
149 containing data (both when sending and receiving), and not of a variable
150 being a pointer to the data. Let's say that I like it better ;)
151 The price for that is constructs like "int msg=*(int*)payload" in
152 handlers, but it's a fine price, IMHO.
154 - Let it work in RL (yuhu)
156 2004-06-21 Martin Quinson
158 - porting SG plugin and SG select to new standards (works almost).
159 - plug memleaks and fix bugs around.
162 - cleanup the prototype of data recv and force users to specify when they
163 want to handle references to objects. Test case working even for cycles.
164 - plug memleaks. Valgrind is perfectly ok with this.
166 2004-06-12 Martin Quinson
168 - cleanup the separation between plugin and main code in plugin creation
170 2004-06-11 Martin Quinson
172 - Reput hook for raw sockets, needed for BW experiments
173 - kill a few lines of dead code
174 [Data description] Interface cleanup
175 - gras_datadesc_by_name returns the searched type or NULL.
176 That way, no variable is needed to use a type desc once, which makes
178 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
179 two parameters were strings (field name, type name), leading to
181 [Dicos] Interface cleanup
182 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
183 This is consistant with the dynar API.
185 2004-04-21 Martin Quinson
187 - Porting to new standards.
190 There is no bag anymore, no need to take extra provision to mask the
191 pointers behind "ID".
192 Better splitup of functions between files create/exchange/convert.
193 This is still a bit artificial since convert and receive are so
194 interleaved, but anyway.
196 - add a queued message list to procdata (the ones not matching criteria
198 - factorize some more code between SG and RL wrt procdata
200 - use gras_exit in example to track memleaks
201 - get rid of gs_example now that GS is properly integrated into gras
202 - update run_test to integrate the lastest tests (datadesc)
204 - rename WARNINGn macros to WARNn since it prooved error-prone
206 2004-04-19 Martin Quinson
208 - register init/exit functions within gras module mecanism
209 - send/receive function.
210 Convertion is not implemented, but short-cutted if not needed.
211 struct/array elements are sent one by one (instead of block-wise), but
212 nobody really cares (yet). Get a prototype before optimizing.
213 - tests (using a file socket) for DD send/receive on:
214 - base types: int, float
215 - array: fixed size, string (ie ref to dynamic string)
216 - structure: homogeneous, heterogeneous
217 - chained list, graph with cycle
218 Believe it or not, valgrind is not too unhappy with the results. The
219 cycle happily segfaults, but the others are ok. And I'm sick of pointers
223 - Bugfix when using a filename explicitely (instead of '-')
225 2004-04-09 Martin Quinson
227 - factorize more code between RL and SG in socket creation
228 - Complete the implementation and tests of:
230 o file (only in RL, and mainly for debugging)
232 I lost 3 days to design a portable address resolver, and then decided
233 that the prototype mainly have to run on my box.
234 Addressing portability too early may be like optimizing too early :-/
236 - use gras_init in the Tests instead of the crappy parse_log_opt
237 (the latter function is removed)
238 [Conditional execution]
239 - New functions: gras_if_RL/gras_if_SG (basic support for this)
240 [Code reorganisation]
241 - Get rid of libgrasutils.a since it makes more trouble than it solves.
242 Build examples against the RL library, since there is no way to disable
243 its creation for now.