Re: [Libcdio-devel] RFC: Support for non-POSIX systems?

[Pete Batard]

On 2012.03.20 08:58, Leon Merten Lohse wrote:
> Already having the code to point to different functions on different
> platforms and having the routines for various OSes it would be a shame
> to support only POSIX.
> How can we call libcdio platform independent if we support only POSIX
> systems and what is the point of this library?
> If we support only POSIX we can get rid of a lot of abstraction that was
> already done.

My point exactly.

> I am a convinced Linux user but imho as long as the projects using
> libcdio like VLC are released for other systems we should try to do so,
> too. The demand is there.

That's also exactly how I see it. Treating MSVC as a minor platform that 
can be brushed aside will do nothing to help the libcdio project, on the 
contrary.

> What we do need are people to test and maintain the releases for the
> different systems. We can talk a lot about what to do or not to do but
> without individual care it will not work out. Every release has to be
> tested on every system we claim to support and should at least have all
> the major features.

Indeed, and, time permitting, I have absolutely no issue being that guy 
for MinGW + MSVC (may even throw cygwin in).

I have been doing so for the past 2 years for libusb (except libusb 
doesn't release, so testing is limited to checking mainline on informal 
milestones), and unless MSVC support is removed from official libcdio, I 
am planning to stick around to perform just that task. Of course, the 
way I see it, MSVC tests will be limited to confirming that it builds, 
and runing ad-hoc checks for the couple of test MSVC project files that 
are available, as I'm definitely not planning to spend time making MSVC 
testing as comprehensive as other platforms.

> This does not have to be the case with git. If a POSIX developer submits
> a patch imho it would be wrong asking him to test it with MSVC.

This was my point.

It is as ridiculous to request non-POSIX people to check POSIX for every 
singe patch as it is to do the opposite. If they *want* to do it (and we 
can try to ask nicely to see if they can), great, but it shouldn't be a 
requirement.

> I would
> prefer to have somebody make sure MSVC works before a release rather
> than testing every patch for MSVC compatibility.

Again, what I was trying to convey.

Unless one provides something like gerrit + Jenkins, with some form of 
MSVC support, it is unreasonable to expect that every single patch can 
be tested for all major platforms, or that any non-POSIX fix should also 
be tested on POSIX.

The pragmatic approach is to try to get the current repo tested on 
regular basis, as well as every single release, which can be done 
through a release candidate that platform sub-maintainers would greenlight.

> We should encourage people to use the release tarballs. Most of the time
> the git version had some more features but was not as stable and once it
> became stable there was a release.
> If the git version lacks some files MSVC needs, there should at least be
> a guide for MSVC developers of how to generate them.

As long as MSVC support remains in mainline, I'm still planning to 
address that one way or another.

Regards,

/Pete