1 2004-07-21 Martin Quinson
3 - Some documentation cleanups
4 - remove the useless last argument of msgtype_declare
5 - rename the Virtu functions to fit into the 'os' namespace
6 - move headers src/include -> src/include/gras/ and stop fooling with
7 gras -> . symbolic link
9 2004-07-19 Martin Quinson
11 - Build shared library also
12 - Install html doc to the right location
13 - stop removing maintainer files in make clean
14 - build tests only on make check
16 2004-07-13 Martin Quinson
18 - No major issue in previous version => change versionning schema
19 - Re-enable little64 convertion test now that Abdou kindly regenerated the
20 corresponding dataset.
22 2004-07-11 Martin Quinson
24 - Get it working with any kind of structure (we can compute the padding
25 bytes remotely for all the architectures I have access to)
26 - Implement the structure parsing macro (still not quite robust/complete)
27 - Improvement to the remote testing toysuite
29 2004-07-10 Martin Quinson
31 - get ride of a bunch of deprecated macros
32 - actually run the test for two-compliment, not only compile it :-/
33 - test whether the structures get packed (and bail out if yes. Damn.
34 Alignment is a serious matter)
35 - test whether the structures get compacted (but respecting the alignment
36 constraints of each types)
37 - test whether the array fields of structures can straddle alignment boundaries
39 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
40 (took me 2 hours to find that bug, looking at the wrong place)
41 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
42 before sending/receiving and are used to compute the offsets of fields
43 - Given that padding size depend even on compiler options, keep track of
44 alignment and aligned_size only for the current architecture. Not a big
45 deal since we send structure fields one after the other (seems
47 - Add the datastructure used for IEEE paper by the PBIO guys to the test
48 program, let it work on linux/gcc/little32. portability todo.
50 2004-07-08 Martin Quinson
51 - import and improve remote compilation support from FAST
52 - make sure make check works on half a dozen of machines out there
54 2004-07-07 Martin Quinson
55 Let's say it's version 0.0.3 ;)
56 - Implement conversions (yuhu!)
57 - Let it work on solaris (beside conversion, of course)
58 - Stupid me, using rand() to generate the conversion datatests in not wise.
60 2004-07-06 Martin Quinson
61 - Let make dist work, since I'm gonna need it to compile on remote hosts
62 - Let Tests/datadesc_usage write the architecture on which the file was
63 generated as first byte.
64 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
65 architecture descriptions.
66 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
67 on those architectures.
68 - Optimization: send/recv array of scalar in one shoot
70 2004-07-05 Martin Quinson
71 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
73 - Plug a whole bunch of memleaks
74 - each process now have to call gras_{init,exit}. One day, their log
75 settings will be separated
76 - Continue the code factorisation between SG, RL and common in Transport.
78 2004-07-04 Martin Quinson
80 - Redistribution between SG and RL.
81 We wanna have to accept in SG, so move accepted related parts of RL in
82 the common part. (more precisely, the dynar of all known sockets is no
83 more a static in transport.c, but part of the process_data)
86 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
88 2004-07-03 Martin Quinson
89 - Create a new log channel tbx containing dict, set, log, dynar (to shut
90 them all up in one shot)
92 - Fix the ugly case of reference to dynamic array.
93 - New (semi-public) function gras_datadesc_size to allow the messaging
94 layer to malloc the needed space for the buffer.
96 - gras_socket_close now expect the socket to close (and not its address to
97 put NULL in it after it). This is because the socket passed to handlers
98 is one of their argument (=> not writable).
100 - propagate the interface cleanup from last week in datadesc, ie remove a
101 superfluous level of indirection. User pass adress of variable
102 containing data (both when sending and receiving), and not of a variable
103 being a pointer to the data. Let's say that I like it better ;)
104 The price for that is constructs like "int msg=*(int*)payload" in
105 handlers, but it's a fine price, IMHO.
107 - Let it work in RL (yuhu)
109 2004-06-21 Martin Quinson
111 - porting SG plugin and SG select to new standards (works almost).
112 - plug memleaks and fix bugs around.
115 - cleanup the prototype of data recv and force users to specify when they
116 want to handle references to objects. Test case working even for cycles.
117 - plug memleaks. Valgrind is perfectly ok with this.
119 2004-06-12 Martin Quinson
121 - cleanup the separation between plugin and main code in plugin creation
123 2004-06-11 Martin Quinson
125 - Reput hook for raw sockets, needed for BW experiments
126 - kill a few lines of dead code
127 [Data description] Interface cleanup
128 - gras_datadesc_by_name returns the searched type or NULL.
129 That way, no variable is needed to use a type desc once, which makes
131 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
132 two parameters were strings (field name, type name), leading to
134 [Dicos] Interface cleanup
135 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
136 This is consistant with the dynar API.
138 2004-04-21 Martin Quinson
140 - Porting to new standards.
143 There is no bag anymore, no need to take extra provision to mask the
144 pointers behind "ID".
145 Better splitup of functions between files create/exchange/convert.
146 This is still a bit artificial since convert and receive are so
147 interleaved, but anyway.
149 - add a queued message list to procdata (the ones not matching criteria
151 - factorize some more code between SG and RL wrt procdata
153 - use gras_exit in example to track memleaks
154 - get rid of gs_example now that GS is properly integrated into gras
155 - update run_test to integrate the lastest tests (datadesc)
157 - rename WARNINGn macros to WARNn since it prooved error-prone
159 2004-04-19 Martin Quinson
161 - register init/exit functions within gras module mecanism
162 - send/receive function.
163 Convertion is not implemented, but short-cutted if not needed.
164 struct/array elements are sent one by one (instead of block-wise), but
165 nobody really cares (yet). Get a prototype before optimizing.
166 - tests (using a file socket) for DD send/receive on:
167 - base types: int, float
168 - array: fixed size, string (ie ref to dynamic string)
169 - structure: homogeneous, heterogeneous
170 - chained list, graph with cycle
171 Believe it or not, valgrind is not too unhappy with the results. The
172 cycle happily segfaults, but the others are ok. And I'm sick of pointers
176 - Bugfix when using a filename explicitely (instead of '-')
178 2004-04-09 Martin Quinson
180 - factorize more code between RL and SG in socket creation
181 - Complete the implementation and tests of:
183 o file (only in RL, and mainly for debugging)
185 I lost 3 days to design a portable address resolver, and then decided
186 that the prototype mainly have to run on my box.
187 Addressing portability too early may be like optimizing too early :-/
189 - use gras_init in the Tests instead of the crappy parse_log_opt
190 (the latter function is removed)
191 [Conditional execution]
192 - New functions: gras_if_RL/gras_if_SG (basic support for this)
193 [Code reorganisation]
194 - Get rid of libgrasutils.a since it makes more trouble than it solves.
195 Build examples against the RL library, since there is no way to disable
196 its creation for now.