diff options
| author | unknown <cmiller@zippy.cornsilk.net> | 2007-02-13 16:35:29 -0500 |
|---|---|---|
| committer | unknown <cmiller@zippy.cornsilk.net> | 2007-02-13 16:35:29 -0500 |
| commit | 40bf2b37486eac74733367166bad94a05894f9c1 (patch) | |
| tree | 5c1b3bbbf429e5745127941b3eacebf8a66d3fb8 /BitKeeper | |
| parent | 29b63cb15e8d2d85a54790dda6e395fe28c99cb6 (diff) | |
| download | mariadb-git-40bf2b37486eac74733367166bad94a05894f9c1.tar.gz | |
Bug#25807: LOAD DATA INFILE does not work with Pipes
Showstopper and regression against 5.0.24.
Previously, we ignored seek() errors (see Bug#22828) and let seek()s
against pipes fail. Now, since we check that a seek didn't fail,
and return without reading, this bug popped up.
This restores the behavior for file-ish objects that could never be
seek()ed.
mysys/mf_iocache.c:
If we detect early that the file is not tell()able, then we should
assume that it's also not seek()able and therefore we should never
set the (poorly named) "seek_not_done" flag so that we don't immedi-
ately try to seek() when reading later.
The problem was that tell() was returning -1, so when we read later,
we needlessly tried to seek to position (unsigned long) -1 .
Also, if we think we're supposed to seek to a position in a file and
the file is un-tell()able, then abort.
Diffstat (limited to 'BitKeeper')
| -rw-r--r-- | BitKeeper/etc/collapsed | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/BitKeeper/etc/collapsed b/BitKeeper/etc/collapsed index e92ccf9aa73..8dcf561b845 100644 --- a/BitKeeper/etc/collapsed +++ b/BitKeeper/etc/collapsed @@ -39,3 +39,5 @@ 45ba4faf2oqu6eR8fqecR3LfSNcYUg 45ba5238-NKl80QVXzdGo8hO9M75Xg 45c0fdfb2mz6NdOIsLenJtf6_ZelTA +45d1ffcd-r3v8A7uh92hQaMfQM9UPQ +45d21437Vg_-i4uOWyvzYWHESXDP6A |