[oi-dev] May 18th (this year...): Launch date of OpenSXCE x86_64
hugo at myhomeemail.co.uk
hugo at myhomeemail.co.uk
Sat May 11 19:25:42 UTC 2013
Thanks Martin :)
> On 11 May 2013 at 00:43 Martin Bochnig <martin at martux.org> wrote:
>
> Thank you Garrett, all.
> Ok, I don't like to dictate anything.
> Just try and C.
>
> Q:
>
> When will it ship.
>
> A:
>
> In the past I never held a single date, to make such predictions is usually
> difficult.
> This time, though, it is different, because during creation of the SPARC
> version, I modified all pkgdefs in advance always also for i386 and amd64, in
> the same source tree.
> So there is little guessing involved, as I had already compiled almost
> everything on x86 in October (at the then current state), and afterwards
> _always_ keept the pkgdefs in sync by definition, until in had to stop on
> January 30th. For example, before I continued with SPARC in October, I tar'ed
> everything over after the compile on x86.
>
> So the only real task involved: Move my disks containing the devel pool over
> to my prepared DualCore Celeron 2.4GHz (with Sandy bridge), re-build it, just
> "make install -k"'ing the SVR4 packages and also build JDS and its packages,
> the same accordingly for all gates.
> The only actual work is to fix a few packaging build errors , where it fails.
> So on those places where it was difficult to write the pkgdefs for x86
> blindly, while I did this on SPARC.
>
> The 1rst part of the selfhosting work is also completed on SPARC, this
> actually had distracted me last week from continuing with x86.
> Now I identiefied the very files that are missing. However, this is not yet
> reflected by corrected pkgdefs. I think most of us find it more important to
> have something to run and test on x86. So let me suspend selfhosting, although
> only the last mile needs to be gone. All focus shifted to the May 18th release
> of our first x86_64 version of OpenSXCE.
>
> I won't have time to read/answer/write email.
> So the next thing you will hear from me is the download link, on May18th.
>
> p.s. Garrett: I fully understand why you prefer IPS as the basis in Illumos.
> Only to avoid a misunderstanding, OpenSXCE does not use IPS's manifests in
> any way.
> Here again, why:
>
>
> <http://openindiana.org/pipermail/openindiana-discuss/2012-December/010921.html>
>
>
> """
>
> There would have been 2 fundamental approaches: Easy and tougher.
> The first would be to take all IPS manifests and convert them
> semi-automated to SVR4 prototype/pkginfo/depend files.
> And simply to keep all the new pkg sets and pkg names and dependencies
> just intact.
> This would certainly be nice and interesting. Maybe someone want's to
> do that, too :)
> However, this would break dependency resolution backwards
> compatibility to all legacy SVR4 packages in existance.
>
> For this reason, I neglected it.
>
> The second one was, to reverse-engineer everything back to the old times.
> And that's what I did.
>
> Not just for Illumos/OSnet, but for ___all___ consolidations.
> Not just to get old pkgdefs to build without failiure (where they are
> available), but also to re-assign more than 15 thousand (!) unassigned
> files to new pkgdefs (this number for Illumos ON alone!!!)
> Also it was necessary to take older G11N checkouts (just as one
> example), because we need the good old /usr/openwin
> internationalization stuff as well, for OpenXsun.
> """
>
>
>
>
>
> Somebody can try it the other (way easier and million times more convenient)
> way.
> But then this breaks legacy SUNW* deps.
>
> tnx
>
>
> %martin
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openindiana.org/pipermail/oi-dev/attachments/20130511/56cbaacb/attachment-0005.html>
More information about the oi-dev
mailing list