On Oct 28, 2013, at 1:58 PM, Daniel Roßberg wrote:
> 2013/10/28 Christopher Sean Morrison <[email protected]>:
>> I'm usually okay with a source-only releases known to not compile on one or
>> two platforms.
>
> Windows isn't only "one platform". It's widely used, even in the
> Army. Or you could look at the download statistics on sourceforge. A
> colleague has already complained to me about the bad quality of
> BRL-CAD. I said to him that he can not expect to have a compileble
> trunk every day, but a release?
I hear your point. I didn't mean to imply an unimportance of Windows as a
distribution platform in the least. It's our dominant user platform, but lacks
a release maintainer. Until someone steps up to be a maintainer and actually
be responsible for that platform's functionality, I don't think we can make any
guarantee. I think that same holds true for all binary platforms.
We can certainly consider changing our practice, but the current landscape is:
trunk: compiles on at least one platform, no guarantee as to which
RELEASE: no guarantee whatsoever, but is a staging ground for STABLE
STABLE: compiles on at least two distinct platforms, no guarantee as to
which
other branches: no guarantee whatsoever
Trunk and STABLE we're specifically set up that way to encourage frequent
releases (which we've not been good about). By design, that has meant source
releases will generally compile with minimal effort on any platform by someone
knowledgable of that platform.
Currently, posting a new source tarball only requires coordination of two
maintainers or one dev with access to two environments. ANY contributor can
initiate a source release any time so long as they follow the trunk->STABLE
release criteria.
I can certainly appreciate the notion of the project guaranteeing compilation
on a set of "core" platforms, like we used to do, but I think that's the open
source community's collective responsibility. Forcing all core devs to
repeatedly check platforms was abandoned in order to promote a faster
development velocity many years ago, and I think this has largely been
effective and worthwhile.
To date, binary release have been delegated to platform maintainers with myself
handling Mac releases, Jordi handling Linux, Erik for FreeBSD, but nobody for
Windows. Basically, the intent is to eliminate any barriers towards posting a
release and delegate responsibility to as many people as possible. I think
what is needed is for someone to step up and say "I'm willing to be the Windows
platform maintainer", accepting responsibility to sort out Windows build issues
properly and post binary releases as needed.
> Furthermore we are very interested in the latest developments here,
> e.g. the new bounding box features and STEP conversion. To keep on
> track we need the trunk. At least from time to time.
My goal for a very long time now has been that we get to the point where
rolling out binaries just happens automatically because we're doing them so
frequently. Continuously integrating will obviously help with that, but also
just having culture of a fast-changing trunk that is always working and
constantly being posted every few weeks. You shouldn't need to track trunk.
You should be able to track STABLE and have a reasonable expectation that
you'll have the latest and greatest in a couple weeks at most.
Cheers!
Sean
------------------------------------------------------------------------------
Android is increasing in popularity, but the open development platform that
developers love is also attractive to malware creators. Download this white
paper to learn more about secure code signing practices that can help keep
Android apps secure.
http://pubads.g.doubleclick.net/gampad/clk?id=65839951&iu=/4140/ostg.clktrk
_______________________________________________
BRL-CAD Developer mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-devel
