diff options
| author | Manuel Novoa III <mjn3@codepoet.org> | 2001-05-21 16:19:42 +0000 |
|---|---|---|
| committer | Manuel Novoa III <mjn3@codepoet.org> | 2001-05-21 16:19:42 +0000 |
| commit | 8662b5d09896d537141eea24eef5a8f17aea7fc2 (patch) | |
| tree | 6c906dd961afed61213c94ed723ecad7b0bdcc8c /libc | |
| parent | a17d863945361b067988adb12efee72651905845 (diff) | |
| download | uClibc-alpine-8662b5d09896d537141eea24eef5a8f17aea7fc2.tar.bz2 uClibc-alpine-8662b5d09896d537141eea24eef5a8f17aea7fc2.tar.xz | |
Add comment about scanf pushback behavior and glibc difference.
Diffstat (limited to 'libc')
| -rw-r--r-- | libc/stdio/scanf.c | 11 |
1 files changed, 11 insertions, 0 deletions
diff --git a/libc/stdio/scanf.c b/libc/stdio/scanf.c index a8f9f0ed0..dbb92392c 100644 --- a/libc/stdio/scanf.c +++ b/libc/stdio/scanf.c @@ -18,6 +18,17 @@ * fix this, I need to rework the "ungetc" machinery in stdio.c again. * I do have one reference though, that seems to imply scanf has a single * character of lookahead. + * + * May 20, 2001 + * + * Quote from ANSI/ISO C99 standard: + * + * fscanf pushes back at most one input character onto the input stream. + * Therefore, some sequences that are acceptable to strtod, strtol, etc., + * are unacceptable to fscanf. + * + * So uClibc's *scanf functions conform to the standard, and glibc's + * implementation doesn't for the "100ergs" case mentioned above. */ #include <stdlib.h> |