[OpenIndiana-discuss] OpenIndiana-discuss Digest, Vol 3, Issue 63
hakansom at ohsu.edu
Tue Oct 26 00:38:37 UTC 2010
alasdairrr at gmail.com said:
> We're going the spec files route, as this makes the most sense given the
> options available. As previously mentioned, Guido is working on "OpenIndiana
> Addon Consolidations" to provide additional software for the OS, you can see
> the draft spec here:
I've seen earlier discussions in the list archives about either using
or converting pkgsrc-built software for easy integration. I did not,
however, see anyone mention that Joyent is already doing pkgsrc builds
for their SmartMachine customer systems. There could be some synergy
to take advantage of here, where the pkgsrc community has already
done the work of making things buildable on Solaris/OS/OI, and Joyent
appears to be an active participant in that pkgsrc-based effort.
I have to say that I've used Blastwave, SFW, and TWW (thewrittenword.com)
pre-built packages for Solaris for nearly a decade; I've done compiling
myself for Solaris and other Unix OS's (prior to Linux coming along) for
decades as well, doing it by hand, using RPM spec-files (yes, on Solaris,
really), and the TWW "sb" utilities. I've looked through the pkgbuild
spec-files and tools, and also the various FreeBSD approaches, esp.
the pkgsrc tree and methodology.
Since I have to support multiple platforms at work (university environment),
the solutions which are cross-platform are the ones that I am drawn toward.
For that reason I can't get too excited about the spec files route. It
often happens that folks at our site need some different software (or
version) than is provided by the community, and I'm happy to port it myself
and then contribute that back to a community repository. But if I have
to do it two or three different ways for the same piece of software, because
every platform has chosen a different type of build or packaging system,
I'm going to have a lot less time and energy to contribute to each
community in turn.
I'm not saying the spec-files route is a bad approach; I'm just saying
that if OpenIndiana goes that direction, you won't get as much help from
folks who think and work with cross-platform tools like I do. The
Solaris-based platform is an endangered species here, and as much as I
dearly love it, I simply cannot afford to focus my software-porting
efforts on such a small customer-base.
More information about the OpenIndiana-discuss