1 2004-07-11 Martin Quinson
3 - Get it working with any kind of structure (we can compute the padding
4 bytes remotely for all the architectures I have access to)
5 - Implement the structure parsing macro (still not quite robust/complete)
6 - Improvement to the remote testing toysuite
8 2004-07-10 Martin Quinson
10 - get ride of a bunch of deprecated macros
11 - actually run the test for two-compliment, not only compile it :-/
12 - test whether the structures get packed (and bail out if yes. Damn.
13 Alignment is a serious matter)
14 - test whether the structures get compacted (but respecting the alignment
15 constraints of each types)
16 - test whether the array fields of structures can straddle alignment boundaries
18 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
19 (took me 2 hours to find that bug, looking at the wrong place)
20 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
21 before sending/receiving and are used to compute the offsets of fields
22 - Given that padding size depend even on compiler options, keep track of
23 alignment and aligned_size only for the current architecture. Not a big
24 deal since we send structure fields one after the other (seems
26 - Add the datastructure used for IEEE paper by the PBIO guys to the test
27 program, let it work on linux/gcc/little32. portability todo.
29 2004-07-08 Martin Quinson
30 - import and improve remote compilation support from FAST
31 - make sure make check works on half a dozen of machines out there
33 2004-07-07 Martin Quinson
34 Let's say it's version 0.0.3 ;)
35 - Implement conversions (yuhu!)
36 - Let it work on solaris (beside conversion, of course)
37 - Stupid me, using rand() to generate the conversion datatests in not wise.
39 2004-07-06 Martin Quinson
40 - Let make dist work, since I'm gonna need it to compile on remote hosts
41 - Let Tests/datadesc_usage write the architecture on which the file was
42 generated as first byte.
43 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
44 architecture descriptions.
45 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
46 on those architectures.
47 - Optimization: send/recv array of scalar in one shoot
49 2004-07-05 Martin Quinson
50 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
52 - Plug a whole bunch of memleaks
53 - each process now have to call gras_{init,exit}. One day, their log
54 settings will be separated
55 - Continue the code factorisation between SG, RL and common in Transport.
57 2004-07-04 Martin Quinson
59 - Redistribution between SG and RL.
60 We wanna have to accept in SG, so move accepted related parts of RL in
61 the common part. (more precisely, the dynar of all known sockets is no
62 more a static in transport.c, but part of the process_data)
65 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
67 2004-07-03 Martin Quinson
68 - Create a new log channel tbx containing dict, set, log, dynar (to shut
69 them all up in one shot)
71 - Fix the ugly case of reference to dynamic array.
72 - New (semi-public) function gras_datadesc_size to allow the messaging
73 layer to malloc the needed space for the buffer.
75 - gras_socket_close now expect the socket to close (and not its address to
76 put NULL in it after it). This is because the socket passed to handlers
77 is one of their argument (=> not writable).
79 - propagate the interface cleanup from last week in datadesc, ie remove a
80 superfluous level of indirection. User pass adress of variable
81 containing data (both when sending and receiving), and not of a variable
82 being a pointer to the data. Let's say that I like it better ;)
83 The price for that is constructs like "int msg=*(int*)payload" in
84 handlers, but it's a fine price, IMHO.
86 - Let it work in RL (yuhu)
88 2004-06-21 Martin Quinson
90 - porting SG plugin and SG select to new standards (works almost).
91 - plug memleaks and fix bugs around.
94 - cleanup the prototype of data recv and force users to specify when they
95 want to handle references to objects. Test case working even for cycles.
96 - plug memleaks. Valgrind is perfectly ok with this.
98 2004-06-12 Martin Quinson
100 - cleanup the separation between plugin and main code in plugin creation
102 2004-06-11 Martin Quinson
104 - Reput hook for raw sockets, needed for BW experiments
105 - kill a few lines of dead code
106 [Data description] Interface cleanup
107 - gras_datadesc_by_name returns the searched type or NULL.
108 That way, no variable is needed to use a type desc once, which makes
110 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
111 two parameters were strings (field name, type name), leading to
113 [Dicos] Interface cleanup
114 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
115 This is consistant with the dynar API.
117 2004-04-21 Martin Quinson
119 - Porting to new standards.
122 There is no bag anymore, no need to take extra provision to mask the
123 pointers behind "ID".
124 Better splitup of functions between files create/exchange/convert.
125 This is still a bit artificial since convert and receive are so
126 interleaved, but anyway.
128 - add a queued message list to procdata (the ones not matching criteria
130 - factorize some more code between SG and RL wrt procdata
132 - use gras_exit in example to track memleaks
133 - get rid of gs_example now that GS is properly integrated into gras
134 - update run_test to integrate the lastest tests (datadesc)
136 - rename WARNINGn macros to WARNn since it prooved error-prone
138 2004-04-19 Martin Quinson
140 - register init/exit functions within gras module mecanism
141 - send/receive function.
142 Convertion is not implemented, but short-cutted if not needed.
143 struct/array elements are sent one by one (instead of block-wise), but
144 nobody really cares (yet). Get a prototype before optimizing.
145 - tests (using a file socket) for DD send/receive on:
146 - base types: int, float
147 - array: fixed size, string (ie ref to dynamic string)
148 - structure: homogeneous, heterogeneous
149 - chained list, graph with cycle
150 Believe it or not, valgrind is not too unhappy with the results. The
151 cycle happily segfaults, but the others are ok. And I'm sick of pointers
155 - Bugfix when using a filename explicitely (instead of '-')
157 2004-04-09 Martin Quinson
159 - factorize more code between RL and SG in socket creation
160 - Complete the implementation and tests of:
162 o file (only in RL, and mainly for debugging)
164 I lost 3 days to design a portable address resolver, and then decided
165 that the prototype mainly have to run on my box.
166 Addressing portability too early may be like optimizing too early :-/
168 - use gras_init in the Tests instead of the crappy parse_log_opt
169 (the latter function is removed)
170 [Conditional execution]
171 - New functions: gras_if_RL/gras_if_SG (basic support for this)
172 [Code reorganisation]
173 - Get rid of libgrasutils.a since it makes more trouble than it solves.
174 Build examples against the RL library, since there is no way to disable
175 its creation for now.