Fw: [cgl_discussion] Sources from each build...

Jeremy A. Puhlman jpuhlman at mvista.com
Tue Sep 24 13:30:02 PDT 2002


----- Original Message ----- 
From: "Mika Kukkonen" <mika at osdl.org>
To: "Jeremy A. Puhlman" <jpuhlman at mvista.com>
Cc: <cgl_discussion at osdl.org>; "Tariq Shureih" <tariq.shureih at intel.com>; "RustyLynch" <rusty.lynch at intel.com>
Sent: Tuesday, September 24, 2002 12:49 PM
Subject: Re: Fw: [cgl_discussion] Sources from each build...


> On Tue, 2002-09-24 at 12:20, Jeremy A. Puhlman wrote:
> (...)
> > I know that it was not you that made that decision, but that is
> > just wrong...Simply by compiling the binaries on a distribution you
> > are making them distribution specific...Especially Redhat, since the
> > 7.x series contains a compiler not supported by anyone out side of 
> > Redhat...
> 
> AFAIK, any binaries that exist in developer.osdl.org are meant for
> consumption of our validation subgroup only, i.e. to be used in our
> (= OSDL CGL-WG) testing, and not to be distributed.
> 
> Tariq & Rusty, as there seems to be some confusion about the issue,
> I think you need to work out some way to remove this confusion, possibly
> by moving the binaries into some subdirectory with a README-file saying
> explicitly that the binaries are provided as they are etc.

My point was we don't want binaries...we want sources...In our integration efforts we 
have never used the binaries...In prior engagments with TLT(yes I understand this is a 
different project) a very large tarball could be grabbed and pulled down in one foul
swoop...It made life a little easier...
 
> > Also, since all but two distributions use rpm as the package manager,
> > using rpm in no way makes a package Redhat specific...rpm is actually
> > a completely separate project developed and maintained out side
> > of Redhat...
> 
> Again AFAIK, the problem with the RPM is that the spec-files are
> definitely distribution specific.

No less distrobution specific then complined binaries...Which is what is 
currently in the tarballs on the build page...Being less distrobution specific 
was the reasoning given behind changing from rpm/srpms on the build page 
to tarballs, which IMHO incorrect...

> > There will be very few distributions, other then the distribution that
> > the binaries were compiled on, that will actually run those 
> > binaries...
> 
> Once again, binaries are not be run by anybody else but by Rusty's
> people. Certainly you can download them and do whatever you want
> with them, but the only reasonable way for outsiders to access and
> evaluate our work is through the anonymous CVS.

Great...I have no problem with that...We were just used to snapshots of the current 
development sources being avalible via the built page...

> > Open Source, speaks and groks source, not binaries...To be "more open
> > source", the information for accessing cvs anonymously should be right
> > on or right off the main page (especially since it does not follow 
> > the norm)...
> 
> Actually we should have a FAQ on developer site that would contain this
> information plus some other answers. Tariq, want to give it a shot?
> 
> > Like I said I understand you may not be the one that made the 
> > decision, but may be you could pass this along to those
> > on the private list who are making the decisions...(even though a list
> > making decisions on how to be more open source being private doesn't >
> seem very open source IMHO)
> 
> Well, I am the guy who made the decision, or at least blessed it. So if
> you or anybody out there has a complaint and feels that he/she is not
> getting a response, mail me. Or just post it to cgl_discussion, I'll
> pick it up from there in case others do not.

Well I thought that is what I just did...:-)

> What comes to our internal mailing lists, there are several good reasons
> why they are limited to members only, and one of them can be read from
> here: http://www.osdl.org/projects/cgl/cgl_part_agree_qa.html

IIRC MontaVista Software is a member of the cgl group...We are contributing 
significant features... 




More information about the cgl_discussion mailing list