1 2004-07-19 Martin Quinson
3 - Build shared library also
4 - Install html doc to the right location
5 - stop removing maintainer files in make clean
6 - build tests only on make check
8 2004-07-13 Martin Quinson
10 - No major issue in previous version => change versionning schema
11 - Re-enable little64 convertion test now that Abdou kindly regenerated the
12 corresponding dataset.
14 2004-07-11 Martin Quinson
16 - Get it working with any kind of structure (we can compute the padding
17 bytes remotely for all the architectures I have access to)
18 - Implement the structure parsing macro (still not quite robust/complete)
19 - Improvement to the remote testing toysuite
21 2004-07-10 Martin Quinson
23 - get ride of a bunch of deprecated macros
24 - actually run the test for two-compliment, not only compile it :-/
25 - test whether the structures get packed (and bail out if yes. Damn.
26 Alignment is a serious matter)
27 - test whether the structures get compacted (but respecting the alignment
28 constraints of each types)
29 - test whether the array fields of structures can straddle alignment boundaries
31 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
32 (took me 2 hours to find that bug, looking at the wrong place)
33 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
34 before sending/receiving and are used to compute the offsets of fields
35 - Given that padding size depend even on compiler options, keep track of
36 alignment and aligned_size only for the current architecture. Not a big
37 deal since we send structure fields one after the other (seems
39 - Add the datastructure used for IEEE paper by the PBIO guys to the test
40 program, let it work on linux/gcc/little32. portability todo.
42 2004-07-08 Martin Quinson
43 - import and improve remote compilation support from FAST
44 - make sure make check works on half a dozen of machines out there
46 2004-07-07 Martin Quinson
47 Let's say it's version 0.0.3 ;)
48 - Implement conversions (yuhu!)
49 - Let it work on solaris (beside conversion, of course)
50 - Stupid me, using rand() to generate the conversion datatests in not wise.
52 2004-07-06 Martin Quinson
53 - Let make dist work, since I'm gonna need it to compile on remote hosts
54 - Let Tests/datadesc_usage write the architecture on which the file was
55 generated as first byte.
56 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
57 architecture descriptions.
58 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
59 on those architectures.
60 - Optimization: send/recv array of scalar in one shoot
62 2004-07-05 Martin Quinson
63 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
65 - Plug a whole bunch of memleaks
66 - each process now have to call gras_{init,exit}. One day, their log
67 settings will be separated
68 - Continue the code factorisation between SG, RL and common in Transport.
70 2004-07-04 Martin Quinson
72 - Redistribution between SG and RL.
73 We wanna have to accept in SG, so move accepted related parts of RL in
74 the common part. (more precisely, the dynar of all known sockets is no
75 more a static in transport.c, but part of the process_data)
78 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
80 2004-07-03 Martin Quinson
81 - Create a new log channel tbx containing dict, set, log, dynar (to shut
82 them all up in one shot)
84 - Fix the ugly case of reference to dynamic array.
85 - New (semi-public) function gras_datadesc_size to allow the messaging
86 layer to malloc the needed space for the buffer.
88 - gras_socket_close now expect the socket to close (and not its address to
89 put NULL in it after it). This is because the socket passed to handlers
90 is one of their argument (=> not writable).
92 - propagate the interface cleanup from last week in datadesc, ie remove a
93 superfluous level of indirection. User pass adress of variable
94 containing data (both when sending and receiving), and not of a variable
95 being a pointer to the data. Let's say that I like it better ;)
96 The price for that is constructs like "int msg=*(int*)payload" in
97 handlers, but it's a fine price, IMHO.
99 - Let it work in RL (yuhu)
101 2004-06-21 Martin Quinson
103 - porting SG plugin and SG select to new standards (works almost).
104 - plug memleaks and fix bugs around.
107 - cleanup the prototype of data recv and force users to specify when they
108 want to handle references to objects. Test case working even for cycles.
109 - plug memleaks. Valgrind is perfectly ok with this.
111 2004-06-12 Martin Quinson
113 - cleanup the separation between plugin and main code in plugin creation
115 2004-06-11 Martin Quinson
117 - Reput hook for raw sockets, needed for BW experiments
118 - kill a few lines of dead code
119 [Data description] Interface cleanup
120 - gras_datadesc_by_name returns the searched type or NULL.
121 That way, no variable is needed to use a type desc once, which makes
123 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
124 two parameters were strings (field name, type name), leading to
126 [Dicos] Interface cleanup
127 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
128 This is consistant with the dynar API.
130 2004-04-21 Martin Quinson
132 - Porting to new standards.
135 There is no bag anymore, no need to take extra provision to mask the
136 pointers behind "ID".
137 Better splitup of functions between files create/exchange/convert.
138 This is still a bit artificial since convert and receive are so
139 interleaved, but anyway.
141 - add a queued message list to procdata (the ones not matching criteria
143 - factorize some more code between SG and RL wrt procdata
145 - use gras_exit in example to track memleaks
146 - get rid of gs_example now that GS is properly integrated into gras
147 - update run_test to integrate the lastest tests (datadesc)
149 - rename WARNINGn macros to WARNn since it prooved error-prone
151 2004-04-19 Martin Quinson
153 - register init/exit functions within gras module mecanism
154 - send/receive function.
155 Convertion is not implemented, but short-cutted if not needed.
156 struct/array elements are sent one by one (instead of block-wise), but
157 nobody really cares (yet). Get a prototype before optimizing.
158 - tests (using a file socket) for DD send/receive on:
159 - base types: int, float
160 - array: fixed size, string (ie ref to dynamic string)
161 - structure: homogeneous, heterogeneous
162 - chained list, graph with cycle
163 Believe it or not, valgrind is not too unhappy with the results. The
164 cycle happily segfaults, but the others are ok. And I'm sick of pointers
168 - Bugfix when using a filename explicitely (instead of '-')
170 2004-04-09 Martin Quinson
172 - factorize more code between RL and SG in socket creation
173 - Complete the implementation and tests of:
175 o file (only in RL, and mainly for debugging)
177 I lost 3 days to design a portable address resolver, and then decided
178 that the prototype mainly have to run on my box.
179 Addressing portability too early may be like optimizing too early :-/
181 - use gras_init in the Tests instead of the crappy parse_log_opt
182 (the latter function is removed)
183 [Conditional execution]
184 - New functions: gras_if_RL/gras_if_SG (basic support for this)
185 [Code reorganisation]
186 - Get rid of libgrasutils.a since it makes more trouble than it solves.
187 Build examples against the RL library, since there is no way to disable
188 its creation for now.