I have been waiting to catch St. John in the snow for a while, and got my chance today. Sadly, it is not as photogenic as I’d imagined, so I’ll spare everyone the grey building with grey sky interrupted only by white flakes. Instead, some city color with a touch of snow, from a jaunt in the village two weekends ago:
[T]o get attention from us, it is necessary to demonstrate the kind of attitude that leads to competence — alert, thoughtful, observant, willing to be an active partner in developing a solution.–Eric S. Raymond, How To Ask Questions The Smart Way
Like a lot of what ESR used to write, this is remarkably good advice- clearly written, straightforward, and demonstrating a lot of insight into how functional software communities work. Thankfully, lots of people still do as ESR says, not as ESR does. Here is a brief story about one such exchange.
In his recent infamous letter, ESR complained about yum’s performance. Not unreasonable; yum is slow. But we know that ESR is supposedly an expert hacker, and an expert in asking questions. So surely if he’d really meant to get the problem fixed and followed his own advice, his post could have looked something like this:
A bit more on yum performance.
“yum -C search foo” makes 1,536,713 system calls. Even though they’re almost all under a millisecond, that really adds up. You could rewrite it in straight assembler, and it would still be slow. Don’t blame Python, and don’t waste your time rewriting it in C until you figure out what the real cause is.
“apt-cache search foo” makes 1,254 system calls. The only one that takes over a millisecond is the call that reads all 7MB of srcpkgcache.bin. 8ms for that.
yum makes 372,776 reads, 372,115 lseeks, and 474,956 futex calls. apt-cache makes 912 reads and 23 seeks. And remember, apt-cache is dealing with 10x as many packages!
…
All praise strace.
It turns out that when you follow ESR’s advice write something useful and data-laden like this, instead of a publicity-seeking rant, then you get responses that look something like this:
So, after this dialogue Patrick contacted me offlist with some extremely helpful debug information and some code samples to help optimize a number of things.
Not the least of which is the above search method.
For a search of the word ‘foo’ the search went from taking 30s to taking 2s.
It’s a pretty massive improvement and I appreciate the guidance.
It is obviously sort of sad that it took ESR whining so loudly and publicly to encourage someone to do decent profiling work on a key piece of infrastructure; but all it took to get it fixed was one person who stopped whining and instead did the nitty-gritty work of profiling with a widely available tool. The outcome: a faster yum, and a great example of providing useful information and constructive guidance instead of whining.
So here is a great big thanks to all those people who quietly and often without reward do the real work of putting together our software, one email thread at a time. Those are the people who should be getting publicity and our attention, not the losers. Thanks, Patrick, Seth, and everyone else like you guys. May you too someday end up on lwn and /. ;)
(mild tangent, from looking at the jargon file for the first time in ages: how did things from little green footballs make their way into the jargon file? Stole my cultural heritage indeed.)
Some DRM analogies that popped out of my poor brain a while back and never got properly elaborated. These are still mostly in the ‘thinking out loud’ stage, so thoughts/comments/constructive criticisms appreciated, and don’t take them too seriously.
Model T:’any color you want as long as it is black’::ipod:FairPlay
Hypothesis: this works fine now, but people will eventually tire of it and demand more/better. We’re still realistically very early in the digital media game. Alternately, 100 years into the automative age we’re still pretty mediocre at buying on actual quality v. perceived quality.
ipod:DRM skeptics::car:American fans of mass transit
Hypothesis: DRM skeptics, despite being ‘right’ in some senses, are maybe doomed to be ignored by 99% of their compatriots, because utility trumps all other considerations. Possible flaw in the analogy: cars enhance personal autonomy, which overrides their social disutility; ipods diminish personal autonomy, which may at some point override their personal utility.
jobs letter:EU DRM antitrust concerns::MS + Novell agreement:EU patent/format antitrust concerns
Like Microsoft’s agreement with Novell, Jobs’s open letter on DRM was largely about urging European antitrust regulators to look elsewhere, not about actually helping customers. Just as MS tried to make Red Hat and the rest of free software look unreasonable by licensing with Novell, knowing that Novell is not a serious threat to their business, so here is Apple primarily urging the EU to place the blame for DRM’s anti-competitive nature at the feet of the labels, rather than on Apple, knowing that the labels are very unlikely to actually let go of DRM anytime soon.
[Old post that I meant to elaborate on, but never did; seems worth pushing out now; maybe I'll write more about it later if anyone finds it interesting :)
On the day before Christmas, I read a paper on 'spillover' by Brett Frischman and Mark Lemley. In a nutshell, the paper is an attempt to think about the value created when new intellectual property is created, but which for various reasons isn't, can't be, and/or perhaps shouldn't be captured by making IP more like 'real' property. They call this additional value 'spillover'- in copyright, an example would be the period after the expiration of the copyright grant or fair use, both of which represent value which (for various reasons) we don't allocate exclusively to the original copyright owner. [Image of an arguably fair use; search for Koons in the linked article.]
On Christmas itself, Krissa gave me Marks of Opulence, a book about the ties between economic trends and art production. I’ve been plowing through it at high speed, because I’m interested in the links between the financing of art and the financing of free software- both of which are goods which aren’t as predictably monetizable as other similar goods (commercial buildings, for example, or proprietary software.) The book is rife with examples of ‘spillover’- first and foremost the centuries of art created primarily to further the ends of religion, which had unintended secondary beneficial aesthetic effects spilling over hundreds of years, which were not necessarily captured completely by the churches or their sponsors.
I don’t have any stunning revelations from these two pieces yet. But I think there are likely some good nuggets to be had. Spillover is where most of the value in free software is, obviously; we’ve put various labels on it- services, consulting, etc., but I like having one term- ‘spillover’- which covers all of it. Similarly, I’m pleased that there is a very long and distinguished history of great works created whose value was always primarily spillover- there was always, of course, a strong desire to own most of the great works, but other non-proprietary, non-commercial, and often irrational motivations have always played a part in sponsoring these works.
Law links these days invite the writing of essays, not sentences, so there are fewer of them :)
Yargh. For the second time this semester, an excellent, young, FLOSS/CC-involved coder-legal dude is coming to speak at an area law school, and for the second time, I have a conflict I can’t resolve and can’t go. This time it is James Grimmelmann, newly named an associate prof at NYLS, who will participate in a panel at CLS on a very interesting topic- business and legal issues in online games. I’d love to peak in and meet James in the flesh, but no such luck today.
(NB: as usual, for those interested in techlaw occurences in NYC, I have posted this to the InfoLaw NYC calendar. Unfortunately, it wasn’t announced to the general student body here today, so not much warning for those looking at the calendar, either.)
Makes me ill: reading extensively about rape in criminal law last week.
Makes me more ill: reading about attempts to trample on our Constitution, in the name of protecting that same Constitution, in constitutional law this week. At my most generous moments I think they are morons; at my least generous I think they are fascists. Maddening.
[Ed. doh: had intended to finish that before posting. Probably tomorrow..]
Turns out not only were Krissa and I living in sin in Massachusetts, we were actually violating Massachusetts state law.
[Oddly, I learned that during a discussion in Property about the rights of landlords, rather than learning it in my Criminal Law class about rape.]
I intended to blog about this when I saw the FFII news last week, but school got in the way. Anyway, Mark having expanded on it a bit gives me another chance. I’m pleased that Ubuntu/Canonical are getting more active on the patent front, announcing that they’ll actively support Nouveau, and helping create a Free Ubuntu. These are good steps, and I’m glad Ubuntu is taking them. In particular, the idea of directly funding European lobbying on the patent issue is a good one (as distasteful as lobbying is), and I think it would be a great way for other participants in the game to make it clear that they’re on the right page. Similarly, the decision to put off proprietary drivers for six months, while not as permanent/strong as I would hope, gives Nouveau and other such driver projects some more breathing room, and puts a slight bit of additional pressure on Nvidia and ATI to think about following Intel’s lead and free all their drivers.
I’m still uncomfortable with a lot of what Canonical/Ubuntu does in this space (apparent sense that gratis is more important than libre, bragging when moving people from libre tools to gratis ones, lack of formal patent policy a la Fedora) but we should all give them huge thanks and due credit for taking this particular important step in the right direction.
