From f1ed7fea7974218db15155f3b8d2e29d3fe0971c Mon Sep 17 00:00:00 2001 From: Adam Spiers Date: Thu, 11 Apr 2013 13:05:13 +0100 Subject: Documentation: add caveats about I/O buffering for check-{attr,ignore} check-attr and check-ignore have the potential to deadlock callers which do not read back the output in real-time. For example, if a caller writes N paths out and then reads N lines back in, it risks becoming blocked on write() to check-*, and check-* is blocked on write back to the caller. Somebody has to buffer; the pipe buffers provide some leeway, but they are limited. Thanks to Peff for pointing this out: http://article.gmane.org/gmane.comp.version-control.git/220534 Signed-off-by: Adam Spiers Signed-off-by: Junio C Hamano --- Documentation/git.txt | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'Documentation/git.txt') diff --git a/Documentation/git.txt b/Documentation/git.txt index 65c37c4544..122b33f066 100644 --- a/Documentation/git.txt +++ b/Documentation/git.txt @@ -771,8 +771,9 @@ for further details. 'GIT_FLUSH':: If this environment variable is set to "1", then commands such as 'git blame' (in incremental mode), 'git rev-list', 'git log', - and 'git whatchanged' will force a flush of the output stream - after each commit-oriented record have been flushed. If this + 'git check-attr', 'git check-ignore', and 'git whatchanged' will + force a flush of the output stream after each record have been + flushed. If this variable is set to "0", the output of these commands will be done using completely buffered I/O. If this environment variable is not set, git will choose buffered or record-oriented flushing -- cgit v1.2.3