aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRob Landley <rob@landley.net>2019-06-03 18:37:27 -0500
committerRob Landley <rob@landley.net>2019-06-03 18:37:27 -0500
commitfee08385a219b309cf51a7679077e3e9a9a5b906 (patch)
tree72f56cdaeb636cdf434681a26e9ec44e415f7ebe
parentc8a52bd25fac5ccec883362cb7db92cf3a504af5 (diff)
downloadtoybox-fee08385a219b309cf51a7679077e3e9a9a5b906.tar.gz
Remove TAIL_SEEK config option, just always include it.
Yeah, it's twice the size and two codepaths, but seekable is the common case, it won't work in pipelines without a non-seek codepath, and the performance penalty not using seek on large files is a enormous.
-rw-r--r--toys/posix/tail.c9
1 files changed, 1 insertions, 8 deletions
diff --git a/toys/posix/tail.c b/toys/posix/tail.c
index d49a70c2..d7161249 100644
--- a/toys/posix/tail.c
+++ b/toys/posix/tail.c
@@ -20,13 +20,6 @@ config TAIL
-n Output the last NUMBER lines (default 10), +X counts from start
-c Output the last NUMBER bytes, +NUMBER counts from start
-f Follow FILE(s), waiting for more data to be appended
-
-config TAIL_SEEK
- bool "tail seek support"
- default y
- depends on TAIL
- help
- This version uses lseek, which is faster on large files.
*/
#define FOR_tail
@@ -154,7 +147,7 @@ static void do_tail(int fd, char *name)
// The slow codepath is always needed, and can handle all input,
// so make lseek support optional.
- if (CFG_TAIL_SEEK && try_lseek(fd, bytes, lines)) return;
+ if (try_lseek(fd, bytes, lines)) return;
// Read data until we run out, keep a trailing buffer
for (;;) {