------- Comment #3 from bje at gcc dot gnu dot org  2005-11-28 05:22 -------
One problem with this proposal is that sometimes the user just wants one actual
error (the C++ error messages are multiline, but are still one message) and it
is difficult to estimate what a good value of <n> is to ensure that you always
get sufficient context.  When the error information is inadequate, you have to
run the compiler again with a higher value of <n>, which is cumbersome.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20201

Reply via email to