Re: svn commit: r23292 - in trunk: build/generator subversion/libsvn_subr

[David Glasser <gl...@mit.edu>]

On 1/30/07, lgo@tigris.org <lg...@tigris.org> wrote:
> Author: lgo
> Date: Tue Jan 30 11:13:52 2007
> New Revision: 23292
>
> Log:
> Add a define SVN_USE_WIN32_CRASHHANDLER that enables building the crash
> handler, define it by default.

I am not much of an expert in either Windows or build systems, but it
seems like with this patch, the crash handler is always built and
linked in but just not enabled (as a compile-time setting) sometimes.

Naively, it would seem to me that either the choice of whether to use
it should be made at build time and the crash handler not built if it
is not desired, or that the choice should be made at run time (in the
config area or something) and the compile-time directive would just
set a default.  Linking in dead code strikes me as odd.

--dave

-- 
David Glasser | glasser@mit.edu | http://www.davidglasser.net/

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: r23292 - in trunk: build/generator subversion/libsvn_subr

[David Glasser <gl...@mit.edu>]

On 1/30/07, David Glasser <gl...@mit.edu> wrote:
> I am not much of an expert in either Windows or build systems, but it
> seems like with this patch, the crash handler is always built and
> linked in but just not enabled (as a compile-time setting) sometimes.

Lieven points out on IRC that while what I said is technically true,
the file which is built and linked in ends up entirely empty if the
#define is not defined, so no actual code ends up there.  Never mind.

--dave

-- 
David Glasser | glasser@mit.edu | http://www.davidglasser.net/

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org