[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[MiNT] New error codes again



Hi,

I've read all comments on the new error codes and I think I should explain
a little more.

First of all, I test the new error codes for weeks now and I haven't come
across a single application that refuses to work because of the new codes.
This is actually not a big surprise since an error code generally simply
signifies "failure" and there are only few cases where an application
would work around such a failure depending on the particular error code it
got from the kernel.  In such cases I took care (believe me, I've checked
every single change) that applications will still see the error code it
expects.

In the case of Pwaitpid I agree with kellis.  If the new kernel returns
ECHILD instead of the old ENOENT that should be changed back for
compatibility reasons.  The "wrong" ENOENT can easily be changed to an
ECHILD in the libc.

We have also thought about a new domain that activates the new error codes
but it turned out that the overhead resulting from such a concept is not
necessary.  Even very old applications still seem to work.

I would suggest that we experiment a little with the new codes, and then
discuss that again.  I am quite sure that things will mostly work with the
new codes.  But I really think it is acceptable if users now sometimes see
an "unknown error" when on the other hand new applications will report
errors in a more reasonable way.

Ciao

Guido
-- 
http://stud.uni-sb.de/~gufl0000/
mailto:gufl0000@stud.uni-sb.de