I agree that avoiding the multiple build systems would be nice.
Unfortunately, though, windows is an animal unto itself. I would
be hesitant to change the whole build system over to cmake just for the
ideal situation that there be one build system for all platforms. I have
seen projects that maintain the win32 build separate from the *nix builds,
and this does not seem to be too onerous. I'm flexible, but will probably
in the end use the same tools to build as I do now.
One remaining critical issue is the zookeeper_close() call while the client
is in the CONNECTED state. Even in the single-threaded setup this call
blocks when connected instead of using zookeeper_interest() and a callback,
as seems appropriate. The current code can cause an infinite loop. For
this port to be serious, we would need this cleaned up.
On Tue, Aug 31, 2010 at 1:09 PM, Patrick Hunt <phu...@gmail.com> wrote:
> Hi Ben, that's great!. There has been some interest in this, however I'm
> not aware that anyone has done a port.
> Here's how to contrib:
> basically you would create a JIRA and attach your patch against latest
> trunk svn. The committers will review and provide feedback.
> It would be great to not have to manage multiple build systems for the c
> code. Should we switch to something like cmake instead of autotools? Or will
> that work for you (win/cygwin/unix based build I mean).
> On Mon, Aug 30, 2010 at 2:00 PM, Ben Collins <ben.coll...@foundationdb.com
> > wrote:
>> I have a working win32 port of the C API, not depending on Cygwin, that
>> supports the "single-threaded" model of network interaction. It compiles
>> Visual Studio 2010 and works on 64 bit Windows 7. There are know issues,
>> and it is in it's initial stages; but it has been successfully used
>> the java server.
>> I am happy to provide patches, but would like any pointers to efforts
>> already undertaken in this area, or folks to communicate with about this.