Robert Pogson

9 Responses to “Debian Wheezy Is Imminent”

1. 1 MK Mar 8th, 2013 at 11:29 am

   So, which of those numbers made you think the release is imminent? Debian/GNU/RMS/FSF/Linux… It has no schedule, so let us hope they fix all the bugs, and finish writing the name by the end of the year.

   PS: Love it, how you mistake your own experience for reality, again and again and again.

2. 2 Robert Pogson Mar 8th, 2013 at 1:17 pm

   MK wrote, “which of those numbers made you think the release is imminent? “

   Let’s see. M$ ships 50K bugs with Lose ’95, 98 and XP and quibbles over ~50, including a ton of applications, with Debian? HAHA! I am having a good day.

3. 3 DrLoser Mar 8th, 2013 at 3:54 pm

   I doubt that M$ even notices Debian bug reports, Robert, let alone quibbles over them.

   Three things about bugs (from bitter personal experience):

   1. It’s only a bug when you notice it. The present absence of bugs does not guarantee the future absence of bugs.

   2. Nobody cares how buggy a product is, if it does what they want. Here, I will agree with you about Windows 95. Buggy, buggy, buggy. And yet it was the single biggest selling OS in history — because people walked straight past the bugs.

   3. If a piece of software is outside your control, then you have to assume it is bug-ridden. Debian is a packager, no more, no less. All those thousands of packages, let alone the kernel and associated syslibs? Upstream. Not Debian. For all I know, Debian does a terrific job at integrating packages. It may even clock and fix bugs while it’s at it.

   But it’s still just a packager of Upstream’s buggy software. It isn’t Debian’s fault. It’s just the way things work.

4. 4 DrLoser Mar 8th, 2013 at 4:32 pm

   BTW, I am not assuming that the kernel is bug-ridden. It’s not quite as solid as I would like it to be if I were part of the team, but it gets a lot of high-powered attention, and Debian’s policy of staying well behind the arterial gush of the bleeding edge pays off well here.

   And there are a number of packaged items that are fairly solid even in the face of bugs. There are bugs in every C/C++ compiler I have ever come across, free or otherwise, but it’s in the nature of the beast that you can program around them.

   However, that thousands of packages thing? I’m willing to bet that, if you picked one purely at random, I could find at least one quite serious bug in it simply by staring at the code and doing a bit of static analysis.

   That’s the problem with the upstream-downstream model, and it’s not going to go away any time soon.

5. 5 Robert Pogson Mar 8th, 2013 at 11:33 pm

   Dr Loser wrote, “Debian’s policy of staying well behind the arterial gush of the bleeding edge pays off well here.”

   For a couple of months I have been building the latest stable kernel from source a few days after release and I have had few problems. That could be luck that none of my drivers have been smashed but I read the changelogs and find every new release has a few features I care about like RAID or KVM. The Linux kernel seems to be in fine shape. I use Linux 3.8.2 and Debian Wheezy use 3.0. I don’t even have a Debian kernel aboard these days. I usually keep two kernels around the last one and the latest, just in case. I do have fairly old hardware. I think the CPU is AMD Phenom II and the motherboard was new two years ago after a failure.

   It looks to me like Debian has just to fix a few bugs in the installer and a handful of various bugs to release Wheezy. I won’t have to do much to reconfigure my systems as I already use Wheezy. I have it on 6 PCs and a virtual machine or two all working smoothly. One of the extant bugs is the version 7 netboot installer but that is trivially fixed. They just have to build a new/current package instead of one from last summer. Any given weekend could be the new release. By April at the latest, I figure. These guys could throw a party and fix half the bugs and tidy up the next day and go this week.

   At the moment I have a dozen reasons to love Debian and in comparison to Ubuntu GNU/Linux, the most important one is that there is no fearless leader herding us to a cliff… I have a really bloated system with over 2K packages installed and I have no bug-reports to file… Life is good and the first seedlings of 2013 are about to sprout.

6. 6 oiaohm Mar 9th, 2013 at 1:09 am

   DrLoser part of debian maintainers job is to run static analysis on lots of packages. The package build farm runs some automatically. The plan at debian is to make the unavoidable.

   Debian see quality as a work in progress to be always improved. This is more than some Linux distributions out there.

   –That’s the problem with the upstream-downstream model, and it’s not going to go away any time soon.–

   No its a problem with any model. Upstream-downstream makes it no worse if both the upstream and the downstream is working with each other. Debian when a fault is found downstream its reported upstream. You could think of Debian maintainers are way more inspecting alpha/beta testers of code. Different eyes looking at the results see stuff people get into the habit of disregarding.

7. 7 Chad McCullough Mar 9th, 2013 at 1:58 pm

   This is going to be a great release! I’ve been using Wheezy for just under a year (with Xfce) and I’ve experienced no serious problems. And in true Debian fashion, they will release when it’s ready, not when they think everyone wants it to be released. This is one of the many reasons I’ve been a Debian fan for so many years.

8. 8 Ivan Mar 11th, 2013 at 11:15 am

   part of debian maintainers job is to run static analysis on lots of packages.

   And it has such a great history of working out well for all involved, Pete.

9. 9 oiaohm Mar 11th, 2013 at 3:06 pm

   Ivan do you not think Debian changed its maintainer ship rules after that. Yes they did.

   http://wiki.debian.org/buildd

   The process of debian is quite different to back then. Investment in build servers. 2008 no build servers. Running some static analysis the debian maintainer cannot avoid since the build sever will do it.

   Ivan the openssl has gained a testsuite for this. And in fact the testsuite on that is auto run build server side.

   Ivan the way forwards is automation of large sections the QA. Problem is it going to be a fairly long time before that is completely. Reality Open Source on average has more complete test suites than closed source.

   So today maintainer uploading untested patched to build server shows up Ivan really quickly.

   Yes debian has improved system.

   Even so the old openssl did not check for broken random number generator or provide a test for it. So even that debian exposed a flaw that the number openssl was using was meant to be coming from a hardware part. Lack of testing was foolish.

   Microsoft ssl implementation does not test the random number generator before trusting it either.

   You like to point to the bug in debian Ivan it about time you did into exactly what the problem was. Yes debian openssl error was only possible with a bug in upstream that would have come out on defective hardware. Debian just found a very much faster and more dependable way of showing the code base problem.

   Remember what I said about upstream downstream. Ivan openssl developers would have never build openssl the way debian did. So would have never came aware of the flaw in implementation.

   Not all hardware random number generators are created equal. A batch of hardware random number generators defective the same way is possible.

   So in this openssl case the upstream downstream works. Ok not perfectly but it works.

Leave a Reply