<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type"/>
</head><body style="">
<div>
Thanks Martin :)
</div>
<blockquote style="position: relative; border-left: blue 1px solid; padding-left: 10px; margin-left: 0px;" type="cite">
On 11 May 2013 at 00:43 Martin Bochnig <martin@martux.org> wrote:
<br/>
<br/>Thank you Garrett, all.
<br/>Ok, I don't like to dictate anything.
<br/>Just try and C.
<br/>
<br/>Q:
<br/>
<br/>When will it ship.
<br/>
<br/>A:
<br/>
<br/>In the past I never held a single date, to make such predictions is usually difficult.
<br/>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.
<br/>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.
<br/>
<br/>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.
<br/>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.
<br/>
<br/>The 1rst part of the selfhosting work is also completed on SPARC, this actually had distracted me last week from continuing with x86.
<br/>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.
<br/>
<br/>I won't have time to read/answer/write email.
<br/>So the next thing you will hear from me is the download link, on May18th.
<br/>
<br clear="all"/>p.s. Garrett: I fully understand why you prefer IPS as the basis in Illumos.
<br/>Only to avoid a misunderstanding, OpenSXCE does not use IPS's manifests in any way.
<br/>Here again, why:
<br/>
<br/>
<a href="http://openindiana.org/pipermail/openindiana-discuss/2012-December/010921.html">http://openindiana.org/pipermail/openindiana-discuss/2012-December/010921.html</a>
<br/>
<br/>"""
<br/>
<pre>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.
"""<br/><br/><br/>
</pre>
<br/>
<pre>Somebody can try it the other (way easier and million times more convenient) way.<br/>But then this breaks legacy SUNW* deps.
</pre> tnx
<br/>
<br/>
<br/>%martin
<br/>
<br/>
<br/>
<br/>
</blockquote>
<div>
<br/>
</div>
</body></html>