The video showed up on Google [].
At mark 40:20 Raph makes a remark about “assburgers”, which I didn’t quite understand. He says that the person (account) “has assburgers” and that he would score highly on the “assburger metric” [if Advogato’s trust metric had inputs calibrated in assburger units?]. I didn’t think it worthwhile to interrupt him and clear the meaning of the expression.
I’m just back from . It was interesting. Raph reminded about the history of which I started to forget, and also claimed that the new spam-marking mechanism works. I also think that it works (although… mikal?), but also it probably surprises a number of new users who do not realize that they must fill the bio part or else they get deleted immediately.
Greg does not blog the routine, and mostly uses his blog as a platform for Very Important Proclamations. This time it was a proposal to , like it is done for Fedora.
I haven’t taken this in enough to blog about it, but in any case, I don’t have to. It is up to Chuck and David. And the latter .
BTW, one thing…
If the partner misses the release date for the next service pack, their hardware will not be supported within the whole product until 12-18 months later, when the next service pack is released.
In case of RHEL, the aim is to deliver updates about once a quarter. If an IHV comes with some cool feature which has to be supported, like CPUs running in lockstep or whatnot, it typically takes that long. But in case of something that is on track and just misses a release, it’s only a 3 month delay.
All I know about ZFS, I’ve got from cks’s blog.
[]: “you cannot make a storage pool that is a mirror of RAID-Z2 pools.” This seems too bizarre an idea at first. Of course filesystem goes on top of RAID, which is why you can’t do it. Duh! But he has an explanation why he’d want it.
[]: Actually it is a significant issue whenever there’s a bug in controller firmware. Lesson: run JBODs.
I still enjoy the greatest art, but my anime blogging has moved to a secret, undisclosed location. May it flourish there unencumbered by its legacy.
Roland :
However, I still think that using a spinlock around an assignment that’s atomic anyway is at best pretty silly. If you just need a memory barrier, then put an explicit mb() (or wmb() or rmb()) there, along with a fat comment about why you need to mess with memory barriers anyway.
>NOOOOOOOOOOO!
In theory, it’s possible to program in Linux kernel by using nothing but mb(). In practice, expirience teaches us that every use of mb() in drivers is a bug. I’m not kidding here. For some reason, even competent and experienced hackers cannot get it right.
I have no idea why it is so, and I don’t want to go into the ESR’s territory of speculation, “curse of the gifted”, and such. I just want anyone who writes drivers to burn it into their minds that memory barriers are off limits (there are some very specific exceptions for mailboxes on 10GE cards and things like that, but nonetheless).
Keeping people from writing code which blows up when Red Hat ships it and customers run it on bigger boxes than the developer’s laptop is an old hobby of mine. For example, my article about the dangers of read-write locks and spin_lock_bh was written back in 2002, when I blogged at Advogato []. But this barrier business is worse, because it affects more than just novices.
Mr. X writes []:
But I also believe that Merlin costs will drop dramatically as a result of bulk production. Even the smallest Falcon 9 needs ten Merlins in order to achieve orbit (nine on stage 1 and one on stage 2.) Even if only four Falcon 9’s are launched in a given year, it represents forty Merlin engines being produced.
He forgets that F9 is reusable, and thus the number of engines required is lower.
Regarding LJ censorship, I could not bypass an opportunity for testing. Indeed, any comment mentioning DPNI fails (the popular workaround is to write D_P_N_I with underscores, like V_I_A_G_R_A; others exist as well). If LJ decided to drop community/Dpni, I say good for them. Six Apart is not obligated to give D_P_N_I a platform if they fail TOS. However, stuffing their sock into my mouth is uncool. I am a bit concerned that LJ may declare L_i_n_u_x or a_n_i_m_e evil next.
{The result is, of course, that thousands more are going to visit dpni.org and read their mad rants. Those people could turn the hits into money very easily now.}