Boost logo

Boost-Build :

From: David Abrahams (dave_at_[hidden])
Date: 2003-11-25 17:09:14

Return-Path: <jamming-admin_at_[hidden]>
From: Matt Armstrong <matt_at_[hidden]>
To: jamming_at_[hidden]
Message-ID: <20031124170809.GA11596_at_[hidden]>
Mail-Followup-To: jamming_at_[hidden]
Subject: [jamming] [BUG] jam 2.5rc3 and parallel builds
Sender: jamming-admin_at_[hidden]
Errors-To: jamming-admin_at_[hidden]
Precedence: bulk
List-Help: <mailto:jamming-request_at_[hidden]?subject=help>
List-Post: <mailto:jamming_at_[hidden]>
List-Subscribe: <>,
List-Id: Discuss the build tool Jam/MR with other users
List-Unsubscribe: <>,
List-Archive: <>
Date: Mon, 24 Nov 2003 09:08:09 -0800
MIME-Version: 1.0

I've found and isolated a bug with parallel builds (-j<n>) and jam

A small source tree that exhibits the bug is at

A .h file is generated at run time. It gets into the dependency tree
via normal header scanning.

With -j1 builds, the generated.h file is built first and everything is
fine. With -j2 or higher, the generated.h file is built first, but
other .c files that #include it get built in parallel and fail.

I haven't been able to decipher make1[abcd] well enough to figure out
what is going on or develop a fix. I have discovered no workaround
either (e.g. making 'first' depend on generated.h doesn't seem to help).
jamming mailing list - jamming_at_[hidden]


Dave Abrahams
Boost Consulting

Boost-Build list run by bdawes at, david.abrahams at, gregod at, cpdaniel at, john at