2 ** Source tree organization
4 ******************************************************
6 There is at least 5 sub-projects in the tree:
8 - XBT: eXtended Bundle of Tools (low-level toolbox: logging, datatypes).
9 - SURF: a SimUlation aRtiFact. This is the simulation kernel.
10 - MSG: originally MetaSimGrid, MSG is a simple distributed application
12 - GRAS: Grid Reality And Simulation (message passing API with two
13 implementations allowing to compile programs on top of the
14 simulator or for the real life without code modification)
15 - AMOK: Advanced Metacomputing Overlay Kit (high level toolbox; Grid
16 application elements such as distributed database, topology
17 discovery service, and so on)
19 They are all in the same tree because they are complementary tools and
20 having all of them in the same package makes the installation easier
21 for end-users. Moreover, it enables to share the compilation chain and
22 eases the development.
24 The tree is not splited on projects, but on file finality:
25 include/ -> all *public* headers
26 include/xbt/*.h -> one file per module
27 include/gras.h -> file including all modules headers
28 (same for xbt instead of gras)
30 src/Makefile.am -> main makefile. All projects should fit in only one
31 library (I mean 2, RL+SG), which is compiled here.
33 Since all object.o files are placed here, you should
34 choose the name of c files carfully to avoid
37 src/gras/DataDesc -> typical project module
38 src/gras/DataDesc/datadesc_interface.h -> visible to any GRAS modules;
39 masked to the user and GROS/AMOK/SURF
40 src/gras/DataDesc/datadesc_private.h -> visible only from this module
42 So, the modules have 3 levels of publicity for their interface.
43 Private, internal to GRAS, public. Of course, I try to keep as much stuff
46 src/include -> another location for protected headers. Used by SURF, and
47 other should be converted, since this is the Right Thing.
49 testsuite/ -> The more test the better.
50 Same organization than src/ and include/
51 Tests are allowed to load some headers of the module they test.
52 All tests should be listed in run_test.in so that they get
54 They are not listed directly in the check_PROGRAMS part of
55 the makefile because run_test knows about the gras logging
56 features and relaunch with full details the failed tests.
58 examples/ -> Supposed to be copy/pastable by the user, so keep it clear and
59 avoid any kind of trick. In particular, do only include the
62 ** Indentation standard
64 *****************************************************
66 Most files use the Kernighan & Ritchie coding style with 2 spaces of
67 indentation. The indent program can help you to stick to it:
69 indent -kr -br -brs -ce -bbo --dont-break-procedure-type --no-tabs
70 --cuddle-do-while --cuddle-else --indent-level2 --leave-preprocessor-space
71 --no-space-after-function-call-names <myfile>
73 FIXME: this list of arguments is still to be discussed, maybe
76 ** Type naming standard
78 *****************************************************
80 It may sound strange, but the type naming convention was source of intense
81 discution between da GRAS posse members. The convention we came to may not
82 be the best solution, but it has the merit to exist and leave everyone work.
83 So please stick to it.
85 - ???_t is a valid type (builded with typedef)
86 - s_toto_t is a structure (access to fields with .)
87 - s_toto is a structure needing 'struct' keyword to be used
89 - u_toto_t is an union
90 - u_toto is an union needing 'union' keyword to be used
91 - toto_t is an 'object' (struct*)
93 Please to not call toto_t something else than an 'object' (ie, something you
94 have to call _new and _free on it).
97 typedef struct s_toto {} s_toto_t, *toto_t;
98 typedef enum {} e_toto_t;
100 Moreover, only toto_t (and e_toto_t) are public. The rest (mainly s_toto_t)
103 If you see any part of the code not following this convention, this is a
104 bug. Please report it (or fix it yourself if you can).
107 ** Random bits about coding standards and portability
109 *****************************************************
112 Don't use it, or you'll have to check the result (and do some dirty stuff
113 on AIX). Use xbt_malloc (or even better, xbt_new) instead.
116 If possible, avoid size_t and use unsigned long instead. If not,
117 #include <sys/types.h> in all files manipulating size_t
118 do cast it to unsigned long before printing (and use %lu)
120 PRINTF pointer difference
121 printf ("diff = %ld\n", (long) (pointer2 - pointer1));
125 ** Commenting the source: doxygen
127 ****************************************************
129 The global structure of the documentation is in doc/modules.doc
131 The structure of each module (xbt, gras, etc) is in doc/module-<module>.doc
133 The structure of a module is in its public header. This way, you're sure to
134 see all the public interface (and only it). The different parts of the
135 interface are grouped using the @name construct, even if it's buggy. Since
136 parts often get reordered, it's better to add numbers to the parts (so that
137 users can see the intended order).
139 The documentation of each type and macro are also in the public header since
140 this is were they live.
142 The documentation of each function must be in the C file were it lives.
144 Any public element (function, type and macro) must have a @brief part.
147 ** XBT virtualization mecanism
149 ****************************************************
151 There is some functionnalities that we want to virtualize in XBT. We
152 want xbt_time to give the simulated clock when running on top of the
153 simulator, and the host clock when running on a real system. This
154 could be placed in GRAS (and was, historically), but there is some
155 reason to lower it down to XBT.
157 Here is the used naming scheme:
159 - xbt_<module>_<func>(): functions working both in SG and RL
160 - xbt_os_<module>_<func>(): RL functions usable even in simulator
162 That way, in libsimgrid, we still can use native functions if we
163 want to. It may for example be useful to get the real time when
164 implementing the simulator. Think of the SIGINT handler, which
165 wants to see if the user pressed the key twice in a 5 seconds
166 interval. This is of little use to check the simulated time here.
168 Here is the file layout:
170 - xbt_rl_<module>.c: native implementation (xbt_<module>_<func>()).
171 Simply call the corresponding xbt_os_<module>_<func>.
172 Part only of libgras.so
174 - xbt_sg_<module>.c: SIMIX implementation xbt_<module>_<func>()).
175 Simply call the corresponding SIMIX implementation.
176 Part only of libsimgrid.so
178 - xbt_os_<module>.c: body of the functions implementing natively the
179 stuff (xbt_os_<module>_<func>()).
180 Part of both libgras.so and libsimgrid.so
182 Since there is almost nothing in xbt_rl_module.c and xbt_sg_module.c,
183 it'd be better to use symbol aliasing here (to declare in the object
184 code that the same function have two names), but I'm still
185 investigating the portability of the thing to windows.
189 * SimGrid Hacker Survival Guide (FIXME: should be betterly placed)
190 ********************************
192 * If you break the logs (for example while hacking in the dynars), you
193 want to define XBT_LOG_MAYDAY at the beginning of log.h. It will
194 desactivate the whole logging mecanism, switching to printfs
195 instead. SimGrid becomes incredibly verbose when doing so, but it
196 you let you fixing the dynars.