diff options
| author | ajs <ajs> | 2005-04-26 17:14:30 +0000 |
|---|---|---|
| committer | ajs <ajs> | 2005-04-26 17:14:30 +0000 |
| commit | 07334da01ca0a7b77cb11add3102b57077eb03eb (patch) | |
| tree | f1c9c9cc0b4919588e0de7f08b54d81db56c7d67 /lib/buffer.c | |
| parent | 2e7d9ebac1a2698d548fab85bf717b427378c3c1 (diff) | |
2005-04-26 Andrew J. Schorr <ajschorr@alumni.princeton.edu>
* buffer.c (buffer_write): Comment out call to buffer_flush_available.
This should speed up buffering at the expense of a possible increase
in latency in flushing the data if inside a long-running thread.
Diffstat (limited to 'lib/buffer.c')
| -rw-r--r-- | lib/buffer.c | 9 |
1 files changed, 7 insertions, 2 deletions
diff --git a/lib/buffer.c b/lib/buffer.c index 48fd35af..81cd0f2a 100644 --- a/lib/buffer.c +++ b/lib/buffer.c @@ -464,11 +464,16 @@ buffer_write(struct buffer *b, int fd, const void *p, size_t size) { ssize_t nbytes; - /* Attempt to drain the previously buffered data? */ +#if 0 + /* Should we attempt to drain any previously buffered data? This could help + reduce latency in pushing out the data if we are stuck in a long-running + thread that is preventing the main select loop from calling the flush + thread... */ if (b->head && (buffer_flush_available(b, fd) == BUFFER_ERROR)) return BUFFER_ERROR; +#endif if (b->head) - /* Buffer still not empty. */ + /* Buffer is not empty, so do not attempt to write the new data. */ nbytes = 0; else if ((nbytes = write(fd, p, size)) < 0) { |