On Wed, Oct 15, 2003 at 10:27:24AM +0200, Corinna Vinschen wrote: >On Wed, Oct 15, 2003 at 08:50:09AM +0200, Micha Nelissen wrote: >> Pierre A. Humblet wrote: >> >FWIW, wouldn't it be cleaner to make "alternate_charset_active" a >> >member of dev_state instead of introducing a new global variable? >> >> Yes, although original_codepage was a viable candidate for a global >> variable?. Never mind, but then either: > >Well, think "historically grown code". > >> 1) that alternate_charset check which currently is in str_to_con >> (centralized), needs to dispersed over all calls to str_to_con. >> (Currently, 1, AFAICS). Prone to bugs, if you ask me because this could >> be forgotten in the future, unless this one call will remain the only one. >> 2) str_to_con has to become a member of dev_state too. > >Point 2 has some merits. Are you interested to do that change, Micha? >Of course, con_to_str should become a dev_state member then, too.
I think it is crucial that we do this. Otherwise child processes will be out of sync with parents. dev_state is supposed to make sure that doesn't happen. cgf