1 2004-07-13 Martin Quinson
3 - No major issue in previous version => change versionning schema
4 - Re-enable little64 convertion test now that Abdou kindly regenerated the
7 2004-07-11 Martin Quinson
9 - Get it working with any kind of structure (we can compute the padding
10 bytes remotely for all the architectures I have access to)
11 - Implement the structure parsing macro (still not quite robust/complete)
12 - Improvement to the remote testing toysuite
14 2004-07-10 Martin Quinson
16 - get ride of a bunch of deprecated macros
17 - actually run the test for two-compliment, not only compile it :-/
18 - test whether the structures get packed (and bail out if yes. Damn.
19 Alignment is a serious matter)
20 - test whether the structures get compacted (but respecting the alignment
21 constraints of each types)
22 - test whether the array fields of structures can straddle alignment boundaries
24 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
25 (took me 2 hours to find that bug, looking at the wrong place)
26 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
27 before sending/receiving and are used to compute the offsets of fields
28 - Given that padding size depend even on compiler options, keep track of
29 alignment and aligned_size only for the current architecture. Not a big
30 deal since we send structure fields one after the other (seems
32 - Add the datastructure used for IEEE paper by the PBIO guys to the test
33 program, let it work on linux/gcc/little32. portability todo.
35 2004-07-08 Martin Quinson
36 - import and improve remote compilation support from FAST
37 - make sure make check works on half a dozen of machines out there
39 2004-07-07 Martin Quinson
40 Let's say it's version 0.0.3 ;)
41 - Implement conversions (yuhu!)
42 - Let it work on solaris (beside conversion, of course)
43 - Stupid me, using rand() to generate the conversion datatests in not wise.
45 2004-07-06 Martin Quinson
46 - Let make dist work, since I'm gonna need it to compile on remote hosts
47 - Let Tests/datadesc_usage write the architecture on which the file was
48 generated as first byte.
49 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
50 architecture descriptions.
51 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
52 on those architectures.
53 - Optimization: send/recv array of scalar in one shoot
55 2004-07-05 Martin Quinson
56 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
58 - Plug a whole bunch of memleaks
59 - each process now have to call gras_{init,exit}. One day, their log
60 settings will be separated
61 - Continue the code factorisation between SG, RL and common in Transport.
63 2004-07-04 Martin Quinson
65 - Redistribution between SG and RL.
66 We wanna have to accept in SG, so move accepted related parts of RL in
67 the common part. (more precisely, the dynar of all known sockets is no
68 more a static in transport.c, but part of the process_data)
71 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
73 2004-07-03 Martin Quinson
74 - Create a new log channel tbx containing dict, set, log, dynar (to shut
75 them all up in one shot)
77 - Fix the ugly case of reference to dynamic array.
78 - New (semi-public) function gras_datadesc_size to allow the messaging
79 layer to malloc the needed space for the buffer.
81 - gras_socket_close now expect the socket to close (and not its address to
82 put NULL in it after it). This is because the socket passed to handlers
83 is one of their argument (=> not writable).
85 - propagate the interface cleanup from last week in datadesc, ie remove a
86 superfluous level of indirection. User pass adress of variable
87 containing data (both when sending and receiving), and not of a variable
88 being a pointer to the data. Let's say that I like it better ;)
89 The price for that is constructs like "int msg=*(int*)payload" in
90 handlers, but it's a fine price, IMHO.
92 - Let it work in RL (yuhu)
94 2004-06-21 Martin Quinson
96 - porting SG plugin and SG select to new standards (works almost).
97 - plug memleaks and fix bugs around.
100 - cleanup the prototype of data recv and force users to specify when they
101 want to handle references to objects. Test case working even for cycles.
102 - plug memleaks. Valgrind is perfectly ok with this.
104 2004-06-12 Martin Quinson
106 - cleanup the separation between plugin and main code in plugin creation
108 2004-06-11 Martin Quinson
110 - Reput hook for raw sockets, needed for BW experiments
111 - kill a few lines of dead code
112 [Data description] Interface cleanup
113 - gras_datadesc_by_name returns the searched type or NULL.
114 That way, no variable is needed to use a type desc once, which makes
116 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
117 two parameters were strings (field name, type name), leading to
119 [Dicos] Interface cleanup
120 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
121 This is consistant with the dynar API.
123 2004-04-21 Martin Quinson
125 - Porting to new standards.
128 There is no bag anymore, no need to take extra provision to mask the
129 pointers behind "ID".
130 Better splitup of functions between files create/exchange/convert.
131 This is still a bit artificial since convert and receive are so
132 interleaved, but anyway.
134 - add a queued message list to procdata (the ones not matching criteria
136 - factorize some more code between SG and RL wrt procdata
138 - use gras_exit in example to track memleaks
139 - get rid of gs_example now that GS is properly integrated into gras
140 - update run_test to integrate the lastest tests (datadesc)
142 - rename WARNINGn macros to WARNn since it prooved error-prone
144 2004-04-19 Martin Quinson
146 - register init/exit functions within gras module mecanism
147 - send/receive function.
148 Convertion is not implemented, but short-cutted if not needed.
149 struct/array elements are sent one by one (instead of block-wise), but
150 nobody really cares (yet). Get a prototype before optimizing.
151 - tests (using a file socket) for DD send/receive on:
152 - base types: int, float
153 - array: fixed size, string (ie ref to dynamic string)
154 - structure: homogeneous, heterogeneous
155 - chained list, graph with cycle
156 Believe it or not, valgrind is not too unhappy with the results. The
157 cycle happily segfaults, but the others are ok. And I'm sick of pointers
161 - Bugfix when using a filename explicitely (instead of '-')
163 2004-04-09 Martin Quinson
165 - factorize more code between RL and SG in socket creation
166 - Complete the implementation and tests of:
168 o file (only in RL, and mainly for debugging)
170 I lost 3 days to design a portable address resolver, and then decided
171 that the prototype mainly have to run on my box.
172 Addressing portability too early may be like optimizing too early :-/
174 - use gras_init in the Tests instead of the crappy parse_log_opt
175 (the latter function is removed)
176 [Conditional execution]
177 - New functions: gras_if_RL/gras_if_SG (basic support for this)
178 [Code reorganisation]
179 - Get rid of libgrasutils.a since it makes more trouble than it solves.
180 Build examples against the RL library, since there is no way to disable
181 its creation for now.