On 4/26/05, Cliff Woolley <jwoolley@virginia.edu> wrote:
> On Tue, 26 Apr 2005, Joe Orton wrote:
>
> > On Tue, Apr 26, 2005 at 08:09:09AM +0200, Erik Huelsmann wrote:
> > > How about the patch below?
> > >
> > > It could easily be backported, as it could be presented as fixing what
> > > should not have been in APR headers.
> >
> > Will this not break programs on Windows which hitherto presumed that APR
> > defines these types? i.e. it breaks the source-compatibility API
> > guarantee for APR 1.x?
>
> Yes, it would. :(
But do you consider breaking programs which depend on bugs in the APR
implementation a bad thing?
I can only repeat this: the APR api (for which the compat promise is
made) is in the APR_* and apr_* namespaces. These types are outside of
this namespace and turn out to cause problems to achieve the
portability goal. There turns out to be a suitable solution for
replacing it which *does* fall into the given namespaces. So, not only
was it identified as a bug, a solution is also available.
One thing is for sure: as long as pid_t and friends are in the Windows
APR headers, APR is hard to combine with other software which does the
same.
|