🚀 go-pugleaf

In article <4gjd8a$o2e@solutions.solon.com> danpop@mail.cern.ch (Dan
Pop) writes:

|> In <4givf4$lak@solutions.solon.com> kanze@gabi-soft.fr (J. Kanze) writes:

|> :It must be defined as something that getchar cannot return otherwise.

|> Why?

Because the implementation is unusable otherwise:-)?

You are right.  All the standard says is that getchar returns `the next
character (if present) as an unsigned char converted to an int'.  If the
value of the `unsigned char' does not fit into an int, then the
conversion is implementation defined.

This could cause definite problems on a 16 bit machine using 16 bit
char's (Unicode), especially if the implementation decided to do
something exotic for the implementation defined bit.

|> :If char is 32 bits, and all values may really be read from disk, then a
|> :conforming implementation will have to have to have an int that is
|> :larger than 32 bits, in order to handle the above correctly.

|> Or the above has to be rewritten to take into account the possibility
|> that EOF can be the same as a valid character.  Trivial, but not
|> particularly elegant.

Correct, but horrible.

|> :Or else,
|> :it would simply have to bill itself as an unhosted implementation, and
|> :be freed from any requirements in the library area.

|> What requirements in the library area are you talking about?
|> Chapter and verse, please.

I was wrong.
-- 
James Kanze           (+33) 88 14 49 00          email: kanze@gabi-soft.fr
GABI Software, Sarl., 8 rue des Francs Bourgeois, 67000 Strasbourg, France
Conseils, études et réalisations en logiciel orienté objet --
              -- A la recherche d'une activité dans une region francophone