Johannes Schindelin wrote:
Hi, On Thu, 22 Feb 2007, Simon 'corecode' Schubert wrote:Johannes Schindelin wrote:Hi, On Wed, 21 Feb 2007, Nicolas Pitre wrote:On Thu, 22 Feb 2007, Johannes Schindelin wrote:diff --git a/index-pack.c b/index-pack.c index fa9a0e7..5ccf4c4 100644 --- a/index-pack.c +++ b/index-pack.c @@ -457,8 +457,8 @@ static void parse_pack_objects(unsigned char *sha1) /* If input_fd is a file, we should have reached its end now. */ if (fstat(input_fd, &st)) die("cannot fstat packfile: %s", strerror(errno)); - if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes) - die("pack has junk at the end"); + if (input_fd && S_ISREG(st.st_mode) && st.st_size != consumed_bytes) + die("pack has junk at the end: 0%o, %d, %d %d", st.st_mode, (int)st.st_size, (int)consumed_bytes, input_fd); if (!nr_deltas) return;What is this supposed to mean?The funny thing is, if you stream part of the bundle file to index-pack, S_ISREG(st.st_mode) is true, even if input_fd == 0.Well, of course: you opened a regular file and pass this as stdin to index-pack. Maybe something like this would be cleaner: if (IS_REF(st.st_mode) && lseek(input_fd, 0, SEEK_CUR) != st.st_size) die("...");The lseek would fail whenever the input is _not_ a file, dying. Since index-pack is called from fetch-pack, with a socket instead of a file, it would fail for the most common user.
that's why i put IS_REF (should read IS_REG) there. but as nicolas pointed out, this won't work for read-ahead. cheers simon -- Serve - BSD +++ RENT this banner advert +++ ASCII Ribbon /"\ Work - Mac +++ space for low €€€ NOW!1 +++ Campaign \ / Party Enjoy Relax | http://dragonflybsd.org Against HTML \ Dude 2c 2 the max ! http://golden-apple.biz Mail + News / \
Attachment:
signature.asc
Description: OpenPGP digital signature