X-Git-Url: http://info.iut-bm.univ-fcomte.fr/pub/gitweb/simgrid.git/blobdiff_plain/c226af4f5e4ed82d142b9065446658bde792072b..3507aa0f86b4306e340c024cf251349861e53a7a:/doc/gtut-tour-05-globals.doc

diff --git a/doc/gtut-tour-05-globals.doc b/doc/gtut-tour-05-globals.doc
index b35c3f7c63..11ddb12138 100644
--- a/doc/gtut-tour-05-globals.doc
+++ b/doc/gtut-tour-05-globals.doc
@@ -64,6 +64,13 @@ loop:
 \skip while
 \until }
 
+Please note that in our example, only one process creates a global
+structure. But this is naturally completely ok to have several
+processes creating their globals this way. Each of these globals will
+be separated, so process A cannot access globals defined by process B.
+Maybe this implies that the name "globals" is a bit misleading. It
+should be "process state" or something similar.
+
 \section GRAS_tut_tour_callback_pitfall Common pitfall of globals
 
 There is an error that I do myself every other day using globals in GRAS.
@@ -82,6 +89,12 @@ embarass GRAS since it does not have its internal buffer initialized yet,
 and cannot store your data anywhere. That is why doing so triggers an error
 at run time.
 
+Also, as noted above, the gras_userdata_new expects the pointed type,
+not the pointer type. As you can see, in our example, you should pass
+server_data_t to the macro, even if the global variable is later
+defined as being of type server_data_t*.
+
+
 \section GRAS_tut_tour_callback_recap Recaping everything together
 
 The whole program now reads: