tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: individual software releases for third parties

On Fri, 24 Aug 2012 11:17:04 -0500 (CDT)
"Jeremy C. Reed" <> wrote:

> At the beginning of this year, rmind suggested an idea to the Board
> about creating separate packages of TNF softare for the use by
> third parties. Basically this means providing a simple tarball of
> the specific code that can be easily and portably built and installed
> without NetBSD.  rmind suggested some benefits: encourage third
> parties to continue using our code, increase reliance on our code,
> gain more visibility, promote NetBSD trademark, and increase our
> userbase.

This is only my (possibly wrongly justified) concerns:

The NetBSD community not having infinite resources, it could require
more time for developers to ensure that their code works elsewhere,
when they might only have the time to test it on NetBSD itself (or to
improve NetBSD itself).

Bloating currently clean code by a lot of conditionals or linker
tricks/stubs to ensure that it works everywhere.  Perhaps dropping some
security, performance or increasing the footprint at times when the
code achieving these features is too-NetBSD-specific (needs to be
tightly integrated with the kernel).

Would it not be beneficial for NetBSD contractors to be hired by
commercial vendors who want to make use of NetBSD code, and don't those
vendors have more resources to adapt the code for their needs than the
NetBSD developers do?  On the other hand, perhaps a difference here
might be more changes being contributed back, if they know that by
sending patches their maintenance would become NetBSD's responsibility.
Or for them to allocate some of their employees as open source NetBSD
maintainers for some subsystems (more mutually-benefiting work force).

I admit these last concerns might be far fetched: If the NetBSD
userland and/or tools are so easily available for use on any system,
will the PR database fill with Linux+NetBSD or glibc+NetBSD related
bugs (including non-bugs)?  If so, will NetBSD kernel users+developers
care about those?  Will the value of the integrated NetBSD OS decrease
because it's so much easier for those who are only concerned about
hardware support and good userland tools to use them on Linux?

That said, please don't consider this post as an objection.  I
recognize the utility of software that can be shared (NetBSD itself
includes code maintained by other projects obviously), and I think that
your approach about asking for volunteer 6-month frame is a realistic
one, as someone who volunteers will probably also care about handling
the related PRs for instance (one of the above concerns).


Home | Main Index | Thread Index | Old Index