The windows code is now working fine so I have merged the windows branch
in the git repo. If there's no outstanding problems we can make a new
release, I guess 2.2.0 would be a good number given the big addition.
Would it also be a good time to add support for 32 F7 video modes?
(see commits of 2009-4-12 and 2009-6-11) This breaks the API but AFAICT
we would keep backward compatibility.
On Thu, 2012-01-26 at 15:38 +0900, Damien Douxchamps wrote:
> Hi all,
> The list has been a bit quiet lately, so let's stir things up with a big
> news: there's a new Windows stack! To keep a long story short, my
> company needed it so we put a few bucks on the table and Satofumi-san
> did the job. All thanks to him ( ^^)v
> As we had limited time and ressources, we decided to use the CMU 1394
> camera library as a base. The CMU software is LPGL, stable and goes
> around numerous quirks of the Windows platform (32/64bit versions, API
> differences, etc). Also, one can link the library binaries easily,
> without code duplication.
> At this point most things should work, although it's hard to actually
> test everything. If you have a windows computer lying around, please
> give it a try!
> Known limitations:
> - lack of valid time stamps in the dc1394video_frame_t structure. The
> CMU team has problems getting this information from the Windows kernel,
> so the buffer fill time cannot reach us.
> - compilation is only for MinGW/Msys at this time. Let us know if you
> can make libdc1394 compile on MS products (VC?)
> - untested: multi-camera/multi-card capture, hubs, hotplug (is this
> supposed to work anyway?), weird topologies (loops), multi-unit devices
> The source code is located in a new "windows" branch of the GIT
> repository. Temporarily, a tarball is also available on the SF website
> (libdc1394-2.1.4-win.tar.gz) Installation instructions are in the
> dc1394/windows/INSTALL.txt file.
> Have fun!
Cyberdyne Inc, Japan