lyndon@ncc.Nexus.CA (Lyndon Nerenberg) (05/21/88)
Although everyone is talking about sources vs binaries, I wonder
if the real issue isn't escaping us?
Nearly *all* the binary postings I have seen are for one of the
many flavours of PC's. The UNIX sites seem to dislike moving
binaries around, and the PC users have no use for source. To me,
this boils down to a PC vs UNIX argument, not a source vs
binaries one. When was the last time you saw a UNIX binary go by?
Yes, there are a *few*. The ones I'm aware of are for the Sun.
It has been proposed that a top level distribution be set up for
binary postings. I agree with this in principle, however I can
see some problems with it. It still leaves a lot of PC related
traffic under the existing comp.sys.* namespace. The volume will
grow considerably over the next few years, so eliminating only
the binary groups is a short term solution at best.
There is also a growing trend to post other than executable
binary files in the current binaries groups - things like GIF
files come to mind. I like collecting GIF files, however I end
up looking in some unlikely places for them
(comp.sys.binaries.just_about_everything, rec.trek, et al).
Although they are binary files, they aren't "executables" which
is what the binaries groups were originally set up to carry. I
would hate to loose these postings simply because I had no use
for PC executables.
Alright, enough rambling :-) What to do?
I think it's time the net considered the idea of splitting the
group namespace in a way that separates UNIX groups from PC
groups at a much higher level. Rick Adams has commented on how
usenet started as a strictly UNIX related network, and others
have indicated the growing resentment on the part of the "old
guard" towards the growing amount of non-UNIX related traffic.
Splitting the system dependent groups into unix.all and pc.all
would make it much easier for sites to filter out material they
are not interested in. [ NO comments on the proposed names - I'm
simply trying to show a very generic example ] The generic
groups (comp.mail, comp.lang, comp.text, etc.) would remain
within the existing comp.all namespace.
This namespace split would allow the existing backbone to drop
the PC groups if they so desired. This is not to say they would
do so. However, it is not fair to expect a backbone (or other)
site to pay a significant amount of money moving groups that are
of no use to them, or their neighbors. I'm sure that the PC sites
don't want to spend money on groups they aren't interested in ei-
ther. Such a split in the group namespace would make it easier
for sites to get the groups they *do* want.
--
{alberta,utzoo,uunet}!ncc!lyndon lyndon@Nexus.CAscott@ksuvax1.cis.ksu.edu (Scott Hammond) (05/23/88)
In article <10226@ncc.Nexus.CA> lyndon@ncc.nexus.ca (Lyndon Nerenberg) writes: >Although everyone is talking about sources vs binaries, I wonder >if the real issue isn't escaping us? > >Nearly *all* the binary postings I have seen are for one of the >many flavours of PC's. The UNIX sites seem to dislike moving >binaries around, and the PC users have no use for source. To me, >this boils down to a PC vs UNIX argument, not a source vs >binaries one.... ... >I think it's time the net considered the idea of splitting the >group namespace in a way that separates UNIX groups from PC >groups at a much higher level. Rick Adams has commented on how >usenet started as a strictly UNIX related network, and others >have indicated the growing resentment on the part of the "old >guard" towards the growing amount of non-UNIX related traffic. >Splitting the system dependent groups into unix.all and pc.all... ... >This namespace split would allow the existing backbone to drop >the PC groups if they so desired... Though our site is almost entirely UNIX oriented, almost every office has a pc or Mac. I personally don't care for binary distributions, but I find it hard to believe so many sites would not benefit from pc related groups. If newsgroups are split by machine type I'm afraid continued introduction of new machines would hopelessly fragment the news (heaven forbid if I should have a pc running UNIX :-)). My concerns about binary distribution have more to do with the risks of running a program I cannot easily study, and which might have unadvertised functions (though I realize a piece of obscure code in a large source distribution could be hard to notice, at least I have a chance). Something else about source vs. binary distribution is that if I see something interesting in source for a pc, I can always attempt to port it to whatever I'm using... I might learn something from an ibmpc discussion, or by studying someone else's source code. Without a pc, a binary distribution is closed to me, hence, I think, the "growing resentment" you mention. I think the problem is that binaries inherently limit the class of users who could benefit from it, as well as consume what I feel is a disproportionate volume. The USENET is experiencing growing pains. UNIX is no longer the only thing out there, but then is the USENET for the strict benefit of the UNIX community? On the other hand, is it time for the USENET to evolve into "logical subnets" based on propagation of the various groups? Most important, which will best serve the future of the USENET in the face of continued growth? I prefer working within the existing hierarchy. I believe "binaries" as related to executables and pictures properly belongs under the "comp" heading. But thinking about growth, perhaps we will reach a point where users are exchanging digitized audio. Do we want to end up with pc.{ ibm.{exec,sources,d}, amiga.{exec,sources,d} } and images.{gif,pic,etc} and rec.audio.digitized? or bin.{ exec.{ibmpc,amiga,etc}, images.{gif,pix,etc}, audio }? or comp.{ exec.{ibmpc,amiga,etc}, images.{gif,pix,etc} }, and rec.audio.digitized? Something to think about. I hope I've contributed somthing useful to the discussion. -------- Scott Hammond Dept of Computing & Information Sciences scott@ksuvax1.cis.ksu.edu Kansas State University scott@ksuvax1.BITNET attmail!ksuvax1!scott Manhattan, Kansas {cbosgd,pyramid,ucsd}!ncr-sd!ncrwic!ksuvax1!scott (913) 532-6350
donegan@stanton.TCC.COM (Steven P. Donegan) (05/27/88)
I am the proud hacker (old connotation) owner of quite a few diverse systems and would like to put forward my points on binary vs source: 1) source can be (in lots of cases) ported between systems; a definite + 2) binaries can harbor nasty surprises (virus, trojan or other); a definite - 3) source can be very educational 4) binaries almost always are simply used to support the task they were designed for, not anything else. And my final (mediocre) argument, I as a usenet site am not really interested in passing on megabytes of binary which cannot be used by a large portion of my leaf nodes. -- Steven P. Donegan Sr. Telecommunications Analyst Western Digital Corp. donegan@stanton.TCC.COM
peter@ficc.UUCP (Peter da Silva) (06/03/88)
How about a 'binary' distribution, then? Disallow binaries posted to world, and restrict them to the systems that want to carry them. -- -- Peter da Silva, Ferranti International Controls Corporation. -- Phone: 713-274-5180. Remote UUCP: uunet!nuchat!sugar!peter.