Re: valgrind patches, was Re: What's cooking in git.git (Jan 2009, #04; Mon, 19)

From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Jan 20, 2009, 14:50 UTC
Message-ID: <alpine.DEB.1.00.0901201545570.5159@intel-tinevez-2-302>
In-Reply-To: <20090120141932.GB10688@sigill.intra.peff.net>

Hi,

On Tue, 20 Jan 2009, Jeff King wrote:

Show 16 quoted lines

> On Tue, Jan 20, 2009 at 02:51:49PM +0100, Johannes Schindelin wrote:
> 
> > > I think creating it inside the trash directory for each test run 
> > > that wants to use valgrind makes more sense (probably as 
> > > .git/valgrind, which is unlikely to hurt anything but will stay out 
> > > of the way of most of the tests).
> > 
> > Here I disagree.  But I think that test-lib.sh should create it 
> > on-demand, and it should traverse all executables in all paths listed 
> > in $PATH, replacing the ones that start with "git-" ("git" itself 
> > should be the first one) that are no scripts by symlinks to the 
> > valgrind script (which should therefore live in t/), and those that 
> > _are_ scripts by symlinks to $GIT_ROOT/$NAME.
> 
> How will you deal with race conditions between two simultaneously 
> running scripts? I.e., where are you going to put it?

There are no race conditions, as for every git executable, a symbolic link is created, pointing to the valgrind.sh script [*1*].

Besides, what with valgrind being a memory hog, you'd be nuts to call valgrinded scripts simultaneously.

Ciao, Dscho

[*1*] Before anybody complains about symbolic links not being available on Windows, or $GIT_SHELL not being heeded by the valgrind.sh script: get valgrind to compile on those platforms, and _then_ we'll talk again.