Julian Foad wrote:
> Peter N. Lundblad wrote:
>> On Wed, 2 Nov 2005, C. Michael Pilato wrote:
>>
>>> I'm fine with the circular dependency. The error stuff can move into
>>> libsvn_fs, as can the svn_fs_*__canonicalize_abspath(). I caution
>>> against futzing with anything but such low-hanging fruit.
>>
>> How would that work on Windows? Currently, we only have a circular
>> dependency regarding struct definitions. We can't have it on
>> functions (or
>> other symbols). We could do a hack like wrapper_template.h in libsvn_ra,
>> but I like to avoid such if possible.
>
> I'll take that as a declaration that circular dependencies at the
> object code level are not allowed in our Windows build.
That's correct.
> That's fine; it's not a good idea to structure code libraries that way
> anyway.
Oh yes.
> So, that's about the end of it as far as I'm concerned. We can move
> source-level-only things to libsvn_fs, but I don't see any of those at
> the moment. Anything more drastic, given the rest of the comments in
> this thread, I'm not going to consider.
The thing to consider here is that you can share source files between
libsvn_fs_fs and libsvn_fs_base *without* creating a new shared library.
So both libraries will be a bit bigger. Big deal. The overhead of
loading yet another .so is probably much bigger than the overhead of
having a slightly larger .so.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Nov 3 02:18:18 2005