>>> On 9/21/2006 at 1:48 AM, in message <[EMAIL PROTECTED]>, "William A. Rowe, Jr." <[EMAIL PROTECTED]> wrote: > We should verify this is or is not a problem on these platforms, but I have > no way to verify. They both should have profited from my earlier efforts > since both platforms include ../win32/filepath.c - but only running > ./testall > on those two platforms will confirm for certain. > > Joe Orton wrote: >> On Thu, Sep 07, 2006 at 11:58:24PM -0500, William Rowe wrote: >>> I've committed several bugfix changes to path handling for win32, and some >>> other issues are also resolved. The APR_CHECK_FILE must die as well, since >>> it may be at the root of many user reports of pthread-related > synchronization >>> and daemon control flukes on Solaris. >>> >>> Thinking of rolling 0.9/1.2 about next Friday, if that motivates anyone to >>> mop up any other lose ends. >> >> The testnames tests are still failing on Unixes after your changes >> there, I haven't had time to chase that up. Presuming that is harmless >> everything else should be good on 1.2.x. Looks like I never committed >> the license header changes to 0.9.x, hum. >> >> testnames : Line 103: expected <../../..>, but saw <../../../> >> FAILED 1 of 1
The test that you are referring to works fine on NetWare. But I had fix another test that wasn't handling the NetWare variable length volume names. Brad
