1 2004-07-26 Martin Quinson
2 Version 0.3 (protocol not changed; ABI changed)
3 - Major overhault of the datadesc interface to simplify it:
4 - shorted the function names:
5 s/gras_datadesc_declare_struct/gras_datadesc_struct/ and so on
6 - add a trivial way to push/pop integers into the cbps without malloc.
7 This allows to make really generic sub_type description, which simply
8 pop their size of the stack.
9 - add a function gras_datadesc_ref_pop_arr() which does what users want
10 most of the time: Declare a dynamic array (which pops its size of the
11 stack) and declare a reference to it. Poor name, but anyway.
12 - kill the post-send callback, add a post-receive one
14 2004-07-23 Martin Quinson
15 Version 0.2 (protocol changed; ABI changed)
16 - add some testing for cpbs in the test cases, and fix some more bugs.
17 This invalidate again the little64 data file, since I cannot regenerate
19 - remove an awfull optimization in the logging stuff, allowing me to:
21 - learn gcc how to check that the argument match the provided format
22 - fix all errors revealed by gcc after that
23 - internal keys of dict are not \0 terminated. Deal with it properly in
24 loggings instead of segfaulting when the user want to see the logs :-/
26 2004-07-22 Martin Quinson
27 - Fix some stupid bug preventing cbps (callback postit) from working
29 2004-07-21 Martin Quinson
30 - Some documentation cleanups
31 - remove the useless last argument of msgtype_declare
32 - rename the Virtu functions to fit into the 'os' namespace
33 - move headers src/include -> src/include/gras/ and stop fooling with
34 gras -> . symbolic link
35 - make distcheck is now successful
37 2004-07-19 Martin Quinson
39 - Build shared library also
40 - Install html doc to the right location
41 - stop removing maintainer files in make clean
42 - build tests only on make check
44 2004-07-13 Martin Quinson
46 - No major issue in previous version => change versionning schema
47 - Re-enable little64 convertion test now that Abdou kindly regenerated the
48 corresponding dataset.
50 2004-07-11 Martin Quinson
52 - Get it working with any kind of structure (we can compute the padding
53 bytes remotely for all the architectures I have access to)
54 - Implement the structure parsing macro (still not quite robust/complete)
55 - Improvement to the remote testing toysuite
57 2004-07-10 Martin Quinson
59 - get ride of a bunch of deprecated macros
60 - actually run the test for two-compliment, not only compile it :-/
61 - test whether the structures get packed (and bail out if yes. Damn.
62 Alignment is a serious matter)
63 - test whether the structures get compacted (but respecting the alignment
64 constraints of each types)
65 - test whether the array fields of structures can straddle alignment boundaries
67 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
68 (took me 2 hours to find that bug, looking at the wrong place)
69 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
70 before sending/receiving and are used to compute the offsets of fields
71 - Given that padding size depend even on compiler options, keep track of
72 alignment and aligned_size only for the current architecture. Not a big
73 deal since we send structure fields one after the other (seems
75 - Add the datastructure used for IEEE paper by the PBIO guys to the test
76 program, let it work on linux/gcc/little32. portability todo.
78 2004-07-08 Martin Quinson
79 - import and improve remote compilation support from FAST
80 - make sure make check works on half a dozen of machines out there
82 2004-07-07 Martin Quinson
83 Let's say it's version 0.0.3 ;)
84 - Implement conversions (yuhu!)
85 - Let it work on solaris (beside conversion, of course)
86 - Stupid me, using rand() to generate the conversion datatests in not wise.
88 2004-07-06 Martin Quinson
89 - Let make dist work, since I'm gonna need it to compile on remote hosts
90 - Let Tests/datadesc_usage write the architecture on which the file was
91 generated as first byte.
92 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
93 architecture descriptions.
94 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
95 on those architectures.
96 - Optimization: send/recv array of scalar in one shoot
98 2004-07-05 Martin Quinson
99 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
101 - Plug a whole bunch of memleaks
102 - each process now have to call gras_{init,exit}. One day, their log
103 settings will be separated
104 - Continue the code factorisation between SG, RL and common in Transport.
106 2004-07-04 Martin Quinson
108 - Redistribution between SG and RL.
109 We wanna have to accept in SG, so move accepted related parts of RL in
110 the common part. (more precisely, the dynar of all known sockets is no
111 more a static in transport.c, but part of the process_data)
113 [gras_stub_generator]
114 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
116 2004-07-03 Martin Quinson
117 - Create a new log channel tbx containing dict, set, log, dynar (to shut
118 them all up in one shot)
120 - Fix the ugly case of reference to dynamic array.
121 - New (semi-public) function gras_datadesc_size to allow the messaging
122 layer to malloc the needed space for the buffer.
124 - gras_socket_close now expect the socket to close (and not its address to
125 put NULL in it after it). This is because the socket passed to handlers
126 is one of their argument (=> not writable).
128 - propagate the interface cleanup from last week in datadesc, ie remove a
129 superfluous level of indirection. User pass adress of variable
130 containing data (both when sending and receiving), and not of a variable
131 being a pointer to the data. Let's say that I like it better ;)
132 The price for that is constructs like "int msg=*(int*)payload" in
133 handlers, but it's a fine price, IMHO.
135 - Let it work in RL (yuhu)
137 2004-06-21 Martin Quinson
139 - porting SG plugin and SG select to new standards (works almost).
140 - plug memleaks and fix bugs around.
143 - cleanup the prototype of data recv and force users to specify when they
144 want to handle references to objects. Test case working even for cycles.
145 - plug memleaks. Valgrind is perfectly ok with this.
147 2004-06-12 Martin Quinson
149 - cleanup the separation between plugin and main code in plugin creation
151 2004-06-11 Martin Quinson
153 - Reput hook for raw sockets, needed for BW experiments
154 - kill a few lines of dead code
155 [Data description] Interface cleanup
156 - gras_datadesc_by_name returns the searched type or NULL.
157 That way, no variable is needed to use a type desc once, which makes
159 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
160 two parameters were strings (field name, type name), leading to
162 [Dicos] Interface cleanup
163 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
164 This is consistant with the dynar API.
166 2004-04-21 Martin Quinson
168 - Porting to new standards.
171 There is no bag anymore, no need to take extra provision to mask the
172 pointers behind "ID".
173 Better splitup of functions between files create/exchange/convert.
174 This is still a bit artificial since convert and receive are so
175 interleaved, but anyway.
177 - add a queued message list to procdata (the ones not matching criteria
179 - factorize some more code between SG and RL wrt procdata
181 - use gras_exit in example to track memleaks
182 - get rid of gs_example now that GS is properly integrated into gras
183 - update run_test to integrate the lastest tests (datadesc)
185 - rename WARNINGn macros to WARNn since it prooved error-prone
187 2004-04-19 Martin Quinson
189 - register init/exit functions within gras module mecanism
190 - send/receive function.
191 Convertion is not implemented, but short-cutted if not needed.
192 struct/array elements are sent one by one (instead of block-wise), but
193 nobody really cares (yet). Get a prototype before optimizing.
194 - tests (using a file socket) for DD send/receive on:
195 - base types: int, float
196 - array: fixed size, string (ie ref to dynamic string)
197 - structure: homogeneous, heterogeneous
198 - chained list, graph with cycle
199 Believe it or not, valgrind is not too unhappy with the results. The
200 cycle happily segfaults, but the others are ok. And I'm sick of pointers
204 - Bugfix when using a filename explicitely (instead of '-')
206 2004-04-09 Martin Quinson
208 - factorize more code between RL and SG in socket creation
209 - Complete the implementation and tests of:
211 o file (only in RL, and mainly for debugging)
213 I lost 3 days to design a portable address resolver, and then decided
214 that the prototype mainly have to run on my box.
215 Addressing portability too early may be like optimizing too early :-/
217 - use gras_init in the Tests instead of the crappy parse_log_opt
218 (the latter function is removed)
219 [Conditional execution]
220 - New functions: gras_if_RL/gras_if_SG (basic support for this)
221 [Code reorganisation]
222 - Get rid of libgrasutils.a since it makes more trouble than it solves.
223 Build examples against the RL library, since there is no way to disable
224 its creation for now.