Andy Buckley wrote: > Troy Dawson wrote: >> - Ubuntu creep - >> How many labs are seeing this? >> Do they see this as a problem? > > Hi guys, > > I've been meaning to mention something like this for a while, but the > above comment was the eventual catalyst! Unfortunately I won't be at > Hepix, so it'll have to be an online-only comment/discussion. > > I'm not speaking as any sort of lab (or even department) spokesperson, > but the Ubuntu creep issue sounds very relevant to me. In Durham we are > currently trialling Ubuntu on two desktops, with plans to eventually > roll it out as our main desktop deployment. It's already widely used on > staff laptops. We largely get away with this because it's a theory group > and so compatibility with LHC experiment / LCG software (which is often > non-portable because of dumbness rather than any fundamental > incompatibility) isn't an issue. > > The reason for our planned move to Ubuntu (and here I paraphrase, since > I'm not our computing manager) is that while SL is an okay stable > platform for Grid and batch farm nodes, the slow release cycle and > resultant outdated application/desktop software are extremely > frustrating for desktop use. (I also find that ensuring that HEP > software works on both SL and Debian-esque platforms is a good way to > cross-check for subtle bugs.) > > As one of our desktop trialees, I've found Ubuntu (Gutsy, so far) a much > more pleasant experience than SL4, and actually the range of packaged > scientific software is larger. That's (genuinely) not meant as a > criticism of Troy and Connie, who obviously work very diligently on SL, > but 2 people attempting to maintain an entire distro is naturally less > sustainable and dynamic than the whole Debian community + Ubuntu > community + Mark Shuttleworth's semi-bottomless pockets. I'm mindful > here that I don't want you to think of me as a "troll" --- the work > you've done on SL is great, but for me it just can't compete with what > Ubuntu has done for desktop Linux in the last few years.* > > With that in mind, and given that the SL team is so stretched, what > happened to the suggestion (a year or so ago) of moving to a model with > SL as an extra yum/apt repository on top of specified commodity Linux > distributions? The current way looks to me like very hard work by > comparison! > >> Do they see this as a problem 3 or 4 years from now when users are >> still running Ubuntu 7.0.4? > > I know that some of our laptop users have upgraded at least once since > their initial installation. Since Ubuntu's stable releases occur at > predictable times, and the installer alerts suitably privileged desktop > users that an upgrade is possible, I suspect that being stuck on old > versions of Ubuntu is less likely than with SL. But I have no data on that! > > Andy > > [*] To be honest, I also use Ubuntu on my HEP servers now, > (www.hepforge.org, and friends) by preference, so it's not just desktops > and laptops where "Ubuntu creep" is happening. > > -- > Dr Andy Buckley > Institute for Particle Physics Phenomenology > Durham University > 0191 3343798 | 0191 3732613 | www.insectnation.org Hi Andy, Thank you for well thought out reply. You brought up several very good points, and it looks like a very good example of how and why it is happening. Since you are not going to be there, and this is very well written, would you mind if I quoted some or all of this in the Plenary Discussion notes? Thanks Troy -- __________________________________________________ Troy Dawson [log in to unmask] (630)840-6468 Fermilab ComputingDivision/LCSI/CSI DSS Group __________________________________________________