An OS that isn't being used is perfectly stable and unified; when's the last time you've heard of a crash in BeOS?
FWIW, that third-party comment is actually first-party (Lennart Poettering goes by mezcalero on LWN).
As for systemd... I rather like it as a process/services/cgroups manager. I just wish they didn't cram everything else into the same project; I feel (without their extensive implementation experience) that having separated components with stable interfaces between them leads to a better user experience since people can try newer versions of various bits and switch back while bugs get fixed. The important part here being the stable interfaces... and well, Linux userland people, beyond a few notable exceptions like glibc, don't seem to believe in that.
Ah, down thread pointed at the fact that HGST was sold to a mix of WD and Toshiba. Bah. And no edit button.
Then someone will lament how IBM no longer makes drives.
I thought they still do, as HGST (that is, IBM sold the division to Hitachi at some point)?
Miss my DeathStar. Not sure; might have been a 75 GB disk...
Systemd as an init system / process spawning thing is kinda nice, actually. (I'm using it on OpenSUSE; tried Arch very briefly. Used it on Debian/Jessie for a bit because gdm3 needed it to let me login.)
Part of the systemd hate is from things that probably shouldn't live in the same project. People would probably be okay with it as a separate resolverd or something, but... having that coupled to systemd is just strange. One of the strong points of systemd is the ability to start services from a variety of triggers (socket activation, etc.); why can't it be an external project (with the same authors) that gets triggered at the right times? udev, maybe... not sure.
Is there a particular reason to block reading (search) instead of writing, given a highly suspect origin? That is, they can enable search and disable mail/plus/whatever, right?
I guess my question boils down to, what advantage does SEO pieces of shit get from searching Google? The only thing I can think of off the top of my head is to check if their SEOing was successful. That doesn't seem overly useful to me (but then, I've never tried to look at that).
Pretty sure they're running on Androids guts, so kinda?
(I think they took out the UI/Java/whatever layers and are using the Linux kernel that Android uses, plus their own UI layer. See info on Gonk.)
There are two groups of developers here.
Ruby on Rails, the framework, had developers that knew about this general class of vulnerabilities - it's easy to write code that ends up being buggy.
GitHub, the web site (that runs on Rails, and hosts the Rails source repository), knew about the general class of vulnerabilities but not that they had these particular instances of them.
It appears that Homakov tried to get Rails to change the defaults so that these things can't happen unless you ask for them, and was rejected as making the framework more difficult for prototyping use; the opinion on the bug was something along the lines of "the developer using the framework should be protecting against this". He then demonstrated in frustration that this was a bad default, since GitHub is one of the leading sites using the framework and is developed by people generally thought of as knowing what they are doing.
It appears that this has worked and the opinion of the framework developers have changed, and no real damage was done, other than possibly reputation.
GitHub, overall, seemed to be collateral damage.
P.S. I don't think GitHub is open source; Ruby on Rails is.
Do you happen to know how the drive-by PDF exploit manages to keep root, then? I'm curious as I don't see how arbitrary code execution via a PDF vulnerability differs from arbitrary code execution via a cable - what sort of magic allows the former case to bypass the security checks that the latter can't duplicate?
That particular comparison keeps getting reposted as the proof that Theora is feasible.
Theora may or may not be comparable in quality to H.264, but that comparison doesn't tell me either way. It completely ignores the H.264 encoding process, which means that Theora has the advantage of taking however long it needs to compress things. Lots of things involve a time/space (memory or disk) trade off, that needs to be taken into account too.
I don't particularly like the licensing issues around H.264 / MPEG*, but that doesn't mean I am willing to take an unfair comparison either.
(Caveat: I'm a C++ programmer, working on code that has lots of macros.)
The debugger. You can mouseover variables in the source view, and it shows the data (reliably, and points to concrete classes). It lets you switch between threads easily, and shows backtraces you double click on to get to the relevant source code. It uses a normal GUI file browser to let you choose symbols to load, if you haven't set it up beforehand (also via a GUI), and warns when it's out of date. With lots of annoying config file hacking, it can let you display structures in a custom manner.
The closest I've seen on Linux was insight, and that was quite a few years ago (maybe it's improved since?). GDB has a huge barrier to entry, and being line-input based means there's no organization (I don't want my code to be displayed in the same place as my backtrace or my local variables). DDD doesn't reliably display my data, and when it does manage to do so visualizes anything C++ horribly.
I've tried KDevelop (3 and 4) a while back; it absolutely hated dealing with things that has an external build system (i.e. it doesn't work as a pure debugger). Debugging C++ in Eclipse was a joke when I tried it (the one time I did have to work on Java, though, it was pretty nice).
As a reference, I code in Komodo/Eclipse/vim (all on the same code base, depends on what I feel like), on a project that uses autoconf/gmake. That applies to both win32 (via msys+msvc) and Linux. I use MSVC as a pure debugger, not as a code editor.
Yep, that's his personal blog (in fact, explicitly not listed in Planet Mozilla by his choice).
The background is trees - he recently bought a nice wooden house somewhere; there's blog posts about that too.
Odd, your updates should end up in the sandbox (and due to AMO being silly, used to also mean your whole extension ends up on the sandbox, instead of having a last-reviewed version public).
This is of course assuming you haven't been marked as trusted; people who were on AMOv1 were grandfathered in, though I understand that's been mass-removed recently. Other "trusted" authors include google and various mozilla employees, AIUI (but unconfirmed).