On Sat, Mar 29, 2008 at 9:44 AM, Gregory Shimansky
<[EMAIL PROTECTED]> wrote:
> On 29 марта 2008 Nathan Beyer wrote:
> > On Thu, Mar 27, 2008 at 11:46 PM, Alexey Varlamov
> >
> > <[EMAIL PROTECTED]> wrote:
> > > 2008/3/28, Nathan Beyer <[EMAIL PROTECTED]>:
> > > > Is there any specific reason we're still on apr/apr-util 1.2.6?
> > >
> > > I guess the opposite, there were no specific reasons to upgrade so far.
> >
> > True, but it would be nice to eliminate custom patches, if possible.
> >
> > > > I'm trying to get things running on Leopard and it seems 1.2.12 is
> > > > necessity. How do we want to approach such an upgrade?
> > >
> > > Looking through APR patches if they're still actual/compatible and
> > > testing, the more the better - should be enough.
> >
> > Is anyone familiar with the current patches? Any specific tests that
> > might be considered regression tests for those patches?
>
> I created a patch for working_vm/make/resources/patches/apr/misc/win32/env.c.
> The problem was with apr_env_get when the requested variable is defined but
> has no value (empty string). APR used to treat it as an undefined variable.
> The test exists in VM acceptance tests but it requires the system to have a
> variable defined in such way (I encountered the bug on a new windows
> installation with no printer, it had PRINTER variable defined as "").
Was this change made on top of 1.2.6 or a different version? This bit
of code seems to be the major difference between the patch and 1.2.12.
if (size == 0)
/* Mid-air collision?. Somebody must've changed the env. var. */
return APR_INCOMPLETE;
http://svn.apache.org/viewvc/harmony/enhanced/drlvm/trunk/make/resources/patches/apr/misc/win32/env.c?view=markup
This check doesn't exist in 1.2.12.
>
> --
> Gregory
>
