[openbeos] Re: Long state of affairs
- From: "Matt Verran" <matt@xxxxxxxxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Fri, 22 Mar 2002 02:59:34 GMT
Aaron,
ATM I'm not in any fixed location but once settled put me in the team
for testing/source code release.
I do this for Cheltenham & Gloucester Building Society and suggest we
follow a siimilar example eg:-
1) replace/code an item
2) parallel test this item with the r5 counterpart, ensure both are
interchangeable
3) ensure both perform alike in operation
4) mark code as 'fit for outside consumption'
Any queries, give me a shout excuse the delay I'm in the process of
moving.
Regards,
Matt (beshare M)
> I am back!
>
> Let me talk a little bit about what happened, then where we go from
> here.
>
> First off, I got a new machine last weekend (9 days ago) - Dual
> Athlon 1600MP's.
> Beautiful machine, in every way. Works with R5 after the MP patch is
> installed.
>
> Last week I had some issues with app_server crashing. I *suspect*
> that it was some
> bad nastiness in the screensaver that I was running, but I can not be
> sure.
> So, come Thrusday, I built the patch. I installed it, tried the
> preferences apps and
> everything looked really good. I used EasyInstallCreator, since no
> one liked the scripting
> approach that I did for patch 1. I had never used EIC before, but
> everything seemed to
> work OK. So I uploaded the patch and started to send the "here it is"
> message. That was
> when the machine lost its mind. No new app would start. I figured
> that it was some bizarre
> setup thing or something, so I rebooted. At that point, I got a "No
> OS found" message.
>
> Well, knowing too much (i.e. that BFS is in the kernel), I *ASSUMED*
> that the kernel had
> lost its mind, and caused BFS to do bad things. I didn't suspect the
> patch, really, but I was
> being cautious, so I asked Daniel to post to the list and went to bed
> (this was way after 1AM).
> I read my mail from work on Friday, saw the analysis of the patch and
> took it down from the
> SF release area.
>
> Saturday (besides fixing my bathroom sink), I played with this
> machine for hours. I hooked up my
> backup machine (PPro 200), but it didn't like much of anything
> either. Sunday was busy, so I ...
> got sick this morning and worked on it today. I ended up making a new
> BeOS HD (I had a
> spare 15 gig lying around). I then copied home and all of my
> important files from my old HD
> to the new one. I finally added a stock R5 beos dir. Rebooted and it
> worked on my spare machine.
> I patched the kernel (with MP patch) and voila, it works on the dual
> Athelon machine. My bigger,
> normal hard drive is sitting unused in the spare machine.
>
> Because there were so many variables, I did not blame the patch as
> quickly as I should have. That was
> my mistake and I will own up to it.
>
> --------------------------------------------
>
> Now. Since a number of people on the list took the opportunity to
> speak freely, so, too, will I.
>
> I have had a lot of time to think this over (waiting for my MP3's to
> move across an IDE bus, for
> example). Our first patch was not the most successful, by any means.
> Our second patch caused
> (hopefully only me) a fair amount of work. There are certainly
> problems with our process. Part of
> those problems is that I have been doing it all. I am currently MIDI
> team lead, Kernel team lead, USB developer,
> patch coordinator, Project lead, communications with others, and any
> other duty that comes along.
> That is way, way too much. I can not be everywhere doing a good job.
>
> One of the failures of the Be community (and I think that I am just
> as guilty as anyone else, and maybe more
> so) was that we assumed that Be was everything always and that we had
> no responsibility. I got free releases
> from Be, dev support, help on line, etc from Be. I bought 1 R5 Pro CD
> (and that was from Gobe). That doesn't
> seem super equitable. We put Be on the hook for everything. We had no
> responsibilities. Honestly, I think that a
> lot of the failure of Be as a company comes from a lack of good apps.
> That was supposed to be our
> (the communites') job.
>
> In any case. We are now Be *and* the developer community. It is all
> us. OS and testing and apps and support
> and every other thing that needs to be done. It would be very
> dangerous for people to fall back into that old mentality
> of "Let Be do it". It *has* to be "Let's do it". Many people have
> stepped up to the plate in a major way for OBOS.
> But, honestly, not enough. Worse, though, have been those who signed
> up, then disappeared. I don't want to
> single anyone out, but I would guess that there have been around 20
> active contributors to OBOS. Out of hundreds.
> Folks, we can not continue like this. We can not do this alone. Not
> in any reasonable amount of time. OBOS2024 does not
> meet my expectations. :-)
>
> Many folks have complained about the "lack of testing". May I ask you
> how you would supply a patch to people
> for testing? I don't (personally) have a T1. So I use SF's. The
> patches are intended to be tested. I am not sure how people
> would offer them to the "testers". Personally, the only way that I
> know is to make them available for download. SF does
> not offer me a way to make things available to a few (dozen) people.
> So I do what I can. Honestly, the "press" around our
> releases is *not* my idea. People will post and write what they will.
>
> So. If someone out there wants to lead a testing group, please let
> Aaron know. There are already a bazillion people on the "I want to
test" list. If you don't like the organization, help to change it. I
don't have any grandiose master plan. OBOS has grown organically. And
it will continue to do so. It doesn't look much like it did 6 months
ago. And I don't suspect that it will look like it does now, come R1.
>
> >Ok folks, here's the lowdown...
> >
> >The second patch release has been posted on the Sourceforge Files
> > page.
> >It is called 'OpenBeOS-20020315' and you don't want to run it.
> >
> >Let me repeat: DO NOT RUN THIS INSTALL!!!
> >
> >The installer has a serious bug: all the system files are moved from
> >the standard location (/boot/beos/system) on the boot drive. It
> > makes
> >the partition unbootable. The files are all still there -- at /boot/
> >home/Desktop/OpenBeOS/Saved Files/system. Unfortunately, /boot/beos/
> >system is left empty.
> >
> >It appears to be a case of moving files instead of copying them.
> >Whatever the foulup, you cannot boot the partition after running the
> >install and re-starting the machine. Michael Phipps was bitten by
> > this
> >himself and is thus now without a working BeOS system. He tested the
> >installation on his one and only BeOS partition (tisk, tisk) which
> > is
> >now unbootable. He was able to send me an email from his wife's
> >machine, but he has asked me to let everyone know that he will be
> >unavailable thru email for a few days until he has this fixed.
> >Unfortunately, as project admin, he's the only guy with the power to
> >remove the patch file from the Sourceforge page, so we'll have to
> > stare
> >at it for a few days.
> >
> >If you have a second BeOS partition, then it's no big deal to fix
> > the
> >problem. Just mount the troublesome drive while in the other
> > partition
> >and copy all the system files back. For example, the following
> > command
> >line should do it:
> >
> >cp -rf "/BOOT/home/Desktop/OpenBeOS/Saved Files/system" /BOOT/beos
> >
> >only replace 'BOOT' with the real mount name for the volume that the
> >install was run on.
> >
> >If someone downloaded this patch, installed it on their one and only
> >BeOS partition (as Michael did), then they've got a definite
> > problem.
> >If they have a valid R5 CD that can be booted from, great -- just
> > boot
> >from this and do the copy command as above. Michael evidently can't
> > go
> >the CD route because his dual processor machine won't let him (not
> > sure
> >why). People in this circumstance will have to find a way to boot a
> >BeOS partition so they can copy the system files back. One route, if
> >need be, would be to download the BeOS Personal Edition for Windows
> > (or
> >Linux) and use that to boot from. There are probably other, more
> >drastic recovery techniques that are too ugly to mention. I'm not
> > going
> >to worry about this too much yet, tho: I'm not sure that anyone
> >(besides Michael) has been bitten by this. Nobody may have even
> > noticed
> >the new patch download file was there yet anyway.
> >
> >Anyway, that's the scoop.
> >
> >
>
>
>
>
Other related posts:
