Robert A Duff a �crit dans le message : wcc66svllos.fsf@world.std.com... > "Jean-Pierre Rosen" writes: > > > In this example, the question is whether this very simple functionnality was > > worth presenting a fourth variation of strings. > > Well, it wouldn't have been a *fourth* one. If we had chosen to do it > your way, it would have been *instead* of the bounded strings package we > ended up with. Therefore, it would not have added complexity. The > issue was purely whether assignment should fail at run time (as in your > version) versus prevented at compile time (the RM version) -- that > issue, traded off against the extra annoyance of having to instantiate > the thing all over the place. > Once again, bounded strings are great when they are used as a *representation* of some higher level concept. If you instantiate it for the Last_Name of your client data base, you don't want to be able to assign a Last_Name to a Street_Address. Analyses of the problem is required to define what is the proper length for Last_Name, and all Last_Name should have the same maximum length. The issue is different if you see strings as -say- an input buffer. In that case, regular strings, or perhaps Unbounded_Strings, might be more fit. Actually, I don't use my Variable_String that much, except in one case: when playing with annex E, because they fit the remote_types profile, and Unbounded_Strings don't. -- --------------------------------------------------------- J-P. Rosen (Rosen.Adalog@wanadoo.fr) Visit Adalog's web site at http://pro.wanadoo.fr/adalog