1 2004-07-23 Martin Quinson
3 - add some testing for cpbs in the test cases, and fix some more bugs.
4 This invalidate again the little64 data file, since I cannot regenerate
6 - remove an awfull optimization in the logging stuff, allowing me to:
8 - learn gcc how to check that the argument match the provided format
9 - fix all errors revealed by gcc after that
10 - internal keys of dict are not \0 terminated. Deal with it properly in
11 loggings instead of segfaulting when the user want to see the logs :-/
13 2004-07-22 Martin Quinson
14 - Fix some stupid bug preventing cbps (callback postit) from working
16 2004-07-21 Martin Quinson
17 - Some documentation cleanups
18 - remove the useless last argument of msgtype_declare
19 - rename the Virtu functions to fit into the 'os' namespace
20 - move headers src/include -> src/include/gras/ and stop fooling with
21 gras -> . symbolic link
22 - make distcheck is now successful
24 2004-07-19 Martin Quinson
26 - Build shared library also
27 - Install html doc to the right location
28 - stop removing maintainer files in make clean
29 - build tests only on make check
31 2004-07-13 Martin Quinson
33 - No major issue in previous version => change versionning schema
34 - Re-enable little64 convertion test now that Abdou kindly regenerated the
35 corresponding dataset.
37 2004-07-11 Martin Quinson
39 - Get it working with any kind of structure (we can compute the padding
40 bytes remotely for all the architectures I have access to)
41 - Implement the structure parsing macro (still not quite robust/complete)
42 - Improvement to the remote testing toysuite
44 2004-07-10 Martin Quinson
46 - get ride of a bunch of deprecated macros
47 - actually run the test for two-compliment, not only compile it :-/
48 - test whether the structures get packed (and bail out if yes. Damn.
49 Alignment is a serious matter)
50 - test whether the structures get compacted (but respecting the alignment
51 constraints of each types)
52 - test whether the array fields of structures can straddle alignment boundaries
54 - Damnit, double are bigger than float (typo in creation of 'double' datadesc)
55 (took me 2 hours to find that bug, looking at the wrong place)
56 - Add gras_datadesc_declare_{union,struct}_close(). They must be used
57 before sending/receiving and are used to compute the offsets of fields
58 - Given that padding size depend even on compiler options, keep track of
59 alignment and aligned_size only for the current architecture. Not a big
60 deal since we send structure fields one after the other (seems
62 - Add the datastructure used for IEEE paper by the PBIO guys to the test
63 program, let it work on linux/gcc/little32. portability todo.
65 2004-07-08 Martin Quinson
66 - import and improve remote compilation support from FAST
67 - make sure make check works on half a dozen of machines out there
69 2004-07-07 Martin Quinson
70 Let's say it's version 0.0.3 ;)
71 - Implement conversions (yuhu!)
72 - Let it work on solaris (beside conversion, of course)
73 - Stupid me, using rand() to generate the conversion datatests in not wise.
75 2004-07-06 Martin Quinson
76 - Let make dist work, since I'm gonna need it to compile on remote hosts
77 - Let Tests/datadesc_usage write the architecture on which the file was
78 generated as first byte.
79 - Add PowerPC (being also IRIX64), SPARC (also power4) and ALPHA
80 architecture descriptions.
81 - Add datadesc_usage.{i386,ppc,sparc} files being the result of execution
82 on those architectures.
83 - Optimization: send/recv array of scalar in one shoot
85 2004-07-05 Martin Quinson
86 - YEAH! GRAS/SG and GRAS/RL are both able to run the ping example !
88 - Plug a whole bunch of memleaks
89 - each process now have to call gras_{init,exit}. One day, their log
90 settings will be separated
91 - Continue the code factorisation between SG, RL and common in Transport.
93 2004-07-04 Martin Quinson
95 - Redistribution between SG and RL.
96 We wanna have to accept in SG, so move accepted related parts of RL in
97 the common part. (more precisely, the dynar of all known sockets is no
98 more a static in transport.c, but part of the process_data)
100 [gras_stub_generator]
101 - Bug fix: Do call gras_process_init from gras_init (wasnt called in RL).
103 2004-07-03 Martin Quinson
104 - Create a new log channel tbx containing dict, set, log, dynar (to shut
105 them all up in one shot)
107 - Fix the ugly case of reference to dynamic array.
108 - New (semi-public) function gras_datadesc_size to allow the messaging
109 layer to malloc the needed space for the buffer.
111 - gras_socket_close now expect the socket to close (and not its address to
112 put NULL in it after it). This is because the socket passed to handlers
113 is one of their argument (=> not writable).
115 - propagate the interface cleanup from last week in datadesc, ie remove a
116 superfluous level of indirection. User pass adress of variable
117 containing data (both when sending and receiving), and not of a variable
118 being a pointer to the data. Let's say that I like it better ;)
119 The price for that is constructs like "int msg=*(int*)payload" in
120 handlers, but it's a fine price, IMHO.
122 - Let it work in RL (yuhu)
124 2004-06-21 Martin Quinson
126 - porting SG plugin and SG select to new standards (works almost).
127 - plug memleaks and fix bugs around.
130 - cleanup the prototype of data recv and force users to specify when they
131 want to handle references to objects. Test case working even for cycles.
132 - plug memleaks. Valgrind is perfectly ok with this.
134 2004-06-12 Martin Quinson
136 - cleanup the separation between plugin and main code in plugin creation
138 2004-06-11 Martin Quinson
140 - Reput hook for raw sockets, needed for BW experiments
141 - kill a few lines of dead code
142 [Data description] Interface cleanup
143 - gras_datadesc_by_name returns the searched type or NULL.
144 That way, no variable is needed to use a type desc once, which makes
146 - gras_datadesc_declare_[struct|union]_append_name is removed. The last
147 two parameters were strings (field name, type name), leading to
149 [Dicos] Interface cleanup
150 - gras_dico_retrieve -> gras_dico_get ; gras_dico_insert -> gras_dico_set
151 This is consistant with the dynar API.
153 2004-04-21 Martin Quinson
155 - Porting to new standards.
158 There is no bag anymore, no need to take extra provision to mask the
159 pointers behind "ID".
160 Better splitup of functions between files create/exchange/convert.
161 This is still a bit artificial since convert and receive are so
162 interleaved, but anyway.
164 - add a queued message list to procdata (the ones not matching criteria
166 - factorize some more code between SG and RL wrt procdata
168 - use gras_exit in example to track memleaks
169 - get rid of gs_example now that GS is properly integrated into gras
170 - update run_test to integrate the lastest tests (datadesc)
172 - rename WARNINGn macros to WARNn since it prooved error-prone
174 2004-04-19 Martin Quinson
176 - register init/exit functions within gras module mecanism
177 - send/receive function.
178 Convertion is not implemented, but short-cutted if not needed.
179 struct/array elements are sent one by one (instead of block-wise), but
180 nobody really cares (yet). Get a prototype before optimizing.
181 - tests (using a file socket) for DD send/receive on:
182 - base types: int, float
183 - array: fixed size, string (ie ref to dynamic string)
184 - structure: homogeneous, heterogeneous
185 - chained list, graph with cycle
186 Believe it or not, valgrind is not too unhappy with the results. The
187 cycle happily segfaults, but the others are ok. And I'm sick of pointers
191 - Bugfix when using a filename explicitely (instead of '-')
193 2004-04-09 Martin Quinson
195 - factorize more code between RL and SG in socket creation
196 - Complete the implementation and tests of:
198 o file (only in RL, and mainly for debugging)
200 I lost 3 days to design a portable address resolver, and then decided
201 that the prototype mainly have to run on my box.
202 Addressing portability too early may be like optimizing too early :-/
204 - use gras_init in the Tests instead of the crappy parse_log_opt
205 (the latter function is removed)
206 [Conditional execution]
207 - New functions: gras_if_RL/gras_if_SG (basic support for this)
208 [Code reorganisation]
209 - Get rid of libgrasutils.a since it makes more trouble than it solves.
210 Build examples against the RL library, since there is no way to disable
211 its creation for now.