1 2004-07-07 Martin Quinson
2 Let's say it's version 0.0.2 ;)
3 - Implement conversions (yuhu!)
4 - Let it work on solaris (beside conversion, of course)
5 - Stupid me, using rand() to generate the conversion datatests in not wise.
7 2004-07-06 Martin Quinson
8 - Let make dist work, since I'm gonna need it to compile on remote hosts
9 - Let Tests/datadesc_usage write the architecture on which the file was
10 generated as first byte.
11 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
12 architecture descriptions.
13 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
14 on those architectures.
15 - Optimization: send/recv array of scalar in one shoot
17 2004-07-05 Martin Quinson
18 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
20 - Plug a whole bunch of memleaks
21 - each process now have to call gras_{init,exit}. One day, their log
22 settings will be separated
23 - Continue the code factorisation between SG, RL and common in Transport.
25 2004-07-04 Martin Quinson
27 - Redistribution between SG and RL.
28 We wanna have to accept in SG, so move accepted related parts of RL in
29 the common part. (more precisely, the dynar of all known sockets is no
30 more a static in transport.c, but part of the process_data)
33 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
35 2004-07-03 Martin Quinson
36 - Create a new log channel tbx containing dict, set, log, dynar (to shut
37 them all up in one shot)
39 - Fix the ugly case of reference to dynamic array.
40 - New (semi-public) function gras_datadesc_size to allow the messaging
41 layer to malloc the needed space for the buffer.
43 - gras_socket_close now expect the socket to close (and not its address to
44 put NULL in it after it). This is because the socket passed to handlers
45 is one of their argument (=> not writable).
47 - propagate the interface cleanup from last week in datadesc, ie remove a
48 superfluous level of indirection. User pass adress of variable
49 containing data (both when sending and receiving), and not of a variable
50 being a pointer to the data. Let's say that I like it better ;)
51 The price for that is constructs like "int msg=*(int*)payload" in
52 handlers, but it's a fine price, IMHO.
54 - Let it work in RL (yuhu)
56 2004-06-21 Martin Quinson
58 - porting SG plugin and SG select to new standards (works almost).
59 - plug memleaks and fix bugs around.
62 - cleanup the prototype of data recv and force users to specify when they
63 want to handle references to objects. Test case working even for cycles.
64 - plug memleaks. Valgrind is perfectly ok with this.
66 2004-06-12 Martin Quinson
68 - cleanup the separation between plugin and main code in plugin creation
70 2004-06-11 Martin Quinson
72 - Reput hook for raw sockets, needed for BW experiments
73 - kill a few lines of dead code
74 [Data description] Interface cleanup
75 - gras_datadesc_by_name returns the searched type or NULL.
76 That way, no variable is needed to use a type desc once, which makes
78 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
79 two parameters were strings (field name, type name), leading to
81 [Dicos] Interface cleanup
82 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
83 This is consistant with the dynar API.
85 2004-04-21 Martin Quinson
87 - Porting to new standards.
90 There is no bag anymore, no need to take extra provision to mask the
92 Better splitup of functions between files create/exchange/convert.
93 This is still a bit artificial since convert and receive are so
94 interleaved, but anyway.
96 - add a queued message list to procdata (the ones not matching criteria
98 - factorize some more code between SG and RL wrt procdata
100 - use gras_exit in example to track memleaks
101 - get rid of gs_example now that GS is properly integrated into gras
102 - update run_test to integrate the lastest tests (datadesc)
104 - rename WARNINGn macros to WARNn since it prooved error-prone
106 2004-04-19 Martin Quinson
108 - register init/exit functions within gras module mecanism
109 - send/receive function.
110 Convertion is not implemented, but short-cutted if not needed.
111 struct/array elements are sent one by one (instead of block-wise), but
112 nobody really cares (yet). Get a prototype before optimizing.
113 - tests (using a file socket) for DD send/receive on:
114 - base types: int, float
115 - array: fixed size, string (ie ref to dynamic string)
116 - structure: homogeneous, heterogeneous
117 - chained list, graph with cycle
118 Believe it or not, valgrind is not too unhappy with the results. The
119 cycle happily segfaults, but the others are ok. And I'm sick of pointers
123 - Bugfix when using a filename explicitely (instead of '-')
125 2004-04-09 Martin Quinson
127 - factorize more code between RL and SG in socket creation
128 - Complete the implementation and tests of:
130 o file (only in RL, and mainly for debugging)
132 I lost 3 days to design a portable address resolver, and then decided
133 that the prototype mainly have to run on my box.
134 Addressing portability too early may be like optimizing too early :-/
136 - use gras_init in the Tests instead of the crappy parse_log_opt
137 (the latter function is removed)
138 [Conditional execution]
139 - New functions: gras_if_RL/gras_if_SG (basic support for this)
140 [Code reorganisation]
141 - Get rid of libgrasutils.a since it makes more trouble than it solves.
142 Build examples against the RL library, since there is no way to disable
143 its creation for now.