Garrett Rooney wrote:
On 1/20/06, William A. Rowe, Jr. <[EMAIL PROTECTED]> wrote:
If you would like to see 1.2.3 (or 1.3.0) 'corrected' please offer a patch.
If you are asking for a backport to 0.9.8-dev, we would consider that as well
as long as there's no change to ABI and no change to the 'expected' behavior
of the functions. Or, we have to add an ugly apr_foo_ex() which corresponds
to new 1.2 code.
We hope someday soon 0.9 goes away, but with so many 2.0 apache and current
svn installed users, it will take some time, yet.
+1, I'd love to see a backport of this fix, especially if it happens
in time for the next 0.9.x release. We (where 'we' == subversion
developers) are probably going to want to do a release of APR in the
reasonably near future to push out BDB 4.4 support, since Subversion's
starting to grow support for it now, and in order for that to be
releasable we'll need a version of APR that has configure glue that
will find it (which I already committed to APR, it just hasn't made it
in to a release).
-garrett
Sorry that I reply to the "wrong" reply, but I'm reading through gmane,
and I'm not friend with my news reader at the moment.
First, sorry if I sounded off like I thought it was your (as in APRs)
fault of the problem in Subversion. That was not the meaning.
I was more trying to set the record straight as to where the supposed
problem came from (in the code).
I also believe that finding a map between user/group/world and ACLs is
very difficult, perhaps impossible, so no real blame there!
I do *not* think that APR 1.2.2 is in error, but I have to punt the
question if the change in apr_open_file() should be proposed for
backport to the more seasoned APR developers, as I do not know if the
new behavior would need a apr_foo_ex(), or if it would be considered
"just a bug fix" (read: what is the expected behavior?).
Regards,
/Andreas
