# [02:40] <benjoffe> I wonder, would it be good to have a media query
# [02:41] <benjoffe> 'contrast', that is, say a float (0 -- normal) -> (1 - high contrast) that represents the contrast levels the audience is capable of viewing
# [09:16] <zcorpan> "It is basically a formal description of the state-of-the art in the Web platform from 3 years ago." - http://blog.n01se.net/?p=375
# [10:13] <jgraham> benjoffe_: Well Sunday evening is not the busiest time
# [10:14] <benjoffe_> jgraham: Yeah makes sense (I'm GMT+1000 and often forget the reset of the world is far behind)
# [10:15] <jgraham> I expect the people who produce the lowest contrast designs would also be the least likely to use that media query
# [10:15] <jgraham> Also, I can't really imagine browsers shipping with UI to set the user preference
# [10:16] <benjoffe_> jgraham: The idea though is that I would like to provide say red on black links, but it's not good for those with poor eyesight, it would be good if I could selectively make that a different colour only for those who need it
# [10:17] <jgraham> In general I would expect solutions that work independently of the author (e.g. the readability features in various bookmarks and safari) to be bigger wins
# [10:18] <benjoffe_> jgraham: sure features like that are great, but their existence is not enough for me to be able to implement these designs that I'd like to
# [10:20] <benjoffe_> I can see it would be difficult to introduce this kind of media query though, as authors won't implement it until browsers support it and browsers won't support it (or at least emphasise it) until many sites use it
# [10:22] <jgraham> And users won't set the preference unless it affects sites they use
# [10:22] <jgraham> and will be annoyed if they set the pref and still get low-contrast text they can't read
# [10:23] <benjoffe_> It could be a ui button that appears only if the site is making use of the query perhaps (kind of like how safari only shows 'reader' if it passes some heuristic)
# [14:30] <nessy> I'd buy a html5 t-shirt to support the case
# [14:33] <jgraham> MikeSmith_: You mean that no small, open source, project manages to produce a UI as polished as the huge, well funded gmail team in a space that hardly anyone cares about because almost everyoine is happy to just use gmail? What a surprise…
# [16:08] <AryehGregor> I've seen it take hours on occasion. But it's just du that's taking that long, it would take that long regardless.
# [16:08] <AryehGregor> Because it has to look at every directory in the filesystem.
# [16:11] <AryehGregor> btrfs is the cool next-gen filesystem for Linux that's basically a total ripoff of all the good ideas of ZFS, but on an OS that more than three people actually care about.
# [16:11] <MikeSmith_> took 3 minutes 45 seconds on my server filesystem of ~16GB
# [16:12] <AryehGregor> In particular, it encourages you to put everything on one big filesystem, and divide things up into "subvolumes" that you don't have reserve space for in advance. So you can divide the space up into subvolumes for different parts of the filesystem and therefore easily track what's using how much space, quickly remove entire subvolumes, etc., but not have to waste space by deciding how much to reserve for each subvolume in advance.
# [16:13] <AryehGregor> But you can put limits on how big each subvolume is allowed to get, and track how big each one is, and otherwise mostly handle them as independent filesystems, so you get almost all the benefits of separate filesystems.
# [16:13] <AryehGregor> But there are zillions of other cool features, like online resize (shrink as well as grow), online fsck, extremely low-cost snapshotting of subvolumes, improved performance on many workloads, etc. etc. etc.
# [16:23] <espadrine> AryehGregor: By the way, what is "online resize and fsck"? That is one thing I never understood about Btrfs.
# [16:25] <AryehGregor> MikeSmith_, the catch is it's not stable and who knows when it will be stable and until it's stable it will probably lock up and/or eat your data.
# [16:26] <MikeSmith_> that's definitely a bit of a downside
# [16:26] <AryehGregor> Yeah, that's the sort of thing that tends to put a damper on filesystem innovation.
# [16:26] <AryehGregor> If your scheduler or something messes up you can always just switch back to an old kernel and big deal, but filesystems messing up is a bit more of a practical problem.
# [16:27] <espadrine> Well, ZFS has been stable for some time...
# [16:27] <AryehGregor> espadrine, "online resize" means you can increase or decrease the size of the filesystem online, like if you add or remove a disk. ext4 (and ext3 and ext2) can increase filesystem size but not decrease.
# [16:28] <AryehGregor> Online fsck is what it sounds like, btrfs is supposed to be able to check for and repair errors online instead of requiring you to mount your root filesystem read-only and therefore take the the whole system offline if you want to check for or fix errors.
# [16:28] <AryehGregor> ZFS has been stable for some time but still reportedly has reliability issues.
# [16:29] <AryehGregor> I know of more than one site that used ZFS but then switched away because they had problems with it failing randomly.
# [16:29] <AryehGregor> ("more than one site" being toolserver.org, and Wikimedia)
# [16:29] <AryehGregor> Failing randomly or otherwise not living up to expectations.
# [16:29] <jgraham> "Online fsck is what it sounds like" - an excuse for yet more tired /. jokes?
# [16:31] <jgraham> AryehGregor: Isn't there a whole genre of jokes based around the small Levenshtein distance between the word "fsck" and the word "fuck"?
# [16:32] <AryehGregor> I'm not expecting to use it in production anytime within the next few years, though.
# [16:33] <Philip`> If quality depends on testing which depends on number of users, that seems to indicate the quality-improvement process is (at least in part) to wait for users to encounter bugs and lose data and then fix those bugs
# [16:33] <Philip`> Can't they, like, write it correctly in the first place?
# [16:38] <Philip`> It's only a filesystem, and the API is already designed and implemented many times so they don't have to deal with drastically changing requirements - it doesn't seem like a problem that will always grow until it's slightly too hard to implement
# [16:40] <AryehGregor> So you'd think. But in practice, people want filesystems to do all kinds of things, and people think of ever cleverer ways to do them.
# [16:40] <AryehGregor> ZFS and btrfs are a totally different filesystem model from other commonly-used filesystems, they do everything copy-on-write.
# [16:40] <AryehGregor> So there are a ton of practical issues to work out.
# [16:41] <AryehGregor> Like "what happens when you get low on disk space?" Both of them have issues then.
# [16:41] <AryehGregor> btrfs uses some fancy B-tree-based data structure that some guy made up in like 2005.
# [16:43] * Philip` thinks it might be interesting if someone did a book collecting articles from developers of lots of projects, explaining what they see as the fundamentally hard problems they have to face, to show why things are never as easy as they superficially appear
# [16:43] <Philip`> (Most things I've seen before are focusing on solutions, which is generally less interesting than problems)
# [16:44] <jgraham> Probably no one wants to write a book suggesting that they're not clever enough to trivially solve all problems
# [16:45] <Philip`> The developers working on a project have already clearly indicated they can't trivially solve all problems, because if they could then they wouldn't still be working on that project
# [16:45] <Philip`> so the book would be a good way for them to offer excuses for not having finished the project yet
# [16:45] <zewt> i sure don't want to touch a filesystem that hasn't been stable for many years
# [16:46] <zewt> given that it's probably the single most critical piece of code on the entire system
# [16:46] <zewt> i expect i'll be sticking with XFS for a long time to come
# [16:49] <AryehGregor> I'm okay with using a filesystem once it's been the default in major distros for a version or two and there are no widespread reports of breakage.
# [16:49] <AryehGregor> Provided I keep backups of anything important.
# [16:50] <AryehGregor> If a million Ubuntu users haven't hit anything really nasty in a year, my desktop isn't important enough that I'm going to worry about being the first.
# [16:51] * Quits: rimantas (~rimliu@93.93.57.193) (Quit: Leaving)
# [16:51] <zewt> well, the default in debian/ubuntu still tends to be ext3/ext4, but i don't trust that
# [16:51] <AryehGregor> ext3 has been the default filesystem in Linux for like ten years.
# [16:51] <Philip`> Just make sure your backup disks aren't using the new filesystem too
# [16:52] <zewt> if a filesystem defaults to doing a lengthy full fsck every N boots by default, that tells me the developers have very little confidence in it
# [16:52] <espadrine> zewt: I read somewhere that ext developers had decided never to create ext5
# [16:52] <espadrine> they are confident Btrfs will be the future
# [16:52] <Philip`> Maybe they'll go straight from ext4 to ext6
# [16:52] <zewt> it's very strange that distros still use ext*, since xfs is pretty much better in every way
# [16:53] <zewt> for very low-memory systems maybe
# [16:53] <Ms2ger> Obviously ext5 would involve writing a spec for each line of code
# [16:54] <AryehGregor> espadrine, that was decided a few years ago, yes. There was a major Linux filesystem summit, and the plan everyone agreed on was to make short-term easy improvements to ext3 in the form of ext4, and consign the long-term future of Linux filesystems to btrfs.
# [16:56] <espadrine> zewt: XFS looks good, I wonder why I haven't heard of it before
# [16:56] <espadrine> AryehGregor: it cannot be shrunk
# [16:56] <AryehGregor> espadrine, not at all. You can do an extremely fast in-place upgrade of ext4 to btrfs, with rollback.
# [16:57] <AryehGregor> And the difference is mostly transparent, btrfs does everything important ext4 does as far as applications are concerned.
# [16:57] <AryehGregor> So it's just a matter of people waiting till it's stable and reliable.
# [16:57] <zewt> my experience with xfs is that it just works: reliable, very consistently fast, and i havn't had to do an fsck of any kind since it was pre-release
# [16:57] * Quits: CvP (~CvP@123.49.22.41) (Disconnected by services)
# [16:57] <AryehGregor> zewt, my experience with ext4 is the same.
# [16:57] <espadrine> AryehGregor: I meant, people won't want to touch it unless other people use it
# [16:57] <espadrine> which is the issue in Python3
# [16:58] <zewt> ext* fscks by default every N boots (which you can turn off, but if a filesystem does something that hugely expensive by default, I assume it's for a reason)
# [16:58] <AryehGregor> espadrine, to use Python 3 you have to port your code, which is nontrivial for large codebases, and impossible if you rely on unported libraries.
# [16:58] <AryehGregor> Any random person can switch their machine from ext4 to btrfs for the heck of it today if they want.
# [16:59] <AryehGregor> zewt, that could speak to a difference in configuration philosophy more than anything.
# [16:59] <AryehGregor> The only time I've had nontrivial filesystem corruption on ext4 was when the underlying disk was bad.
# [16:59] <Philip`> espadrine: I think the problem with Python 3 is more that it's worse than Python 2 until everybody else in the world has upgraded (so you can be compatible with their applications/libraries), so nobody wants to bother upgrading
# [16:59] <zewt> the philosophy i read into it is "we don't trust our code enough"; others may read what they like, of course
# [16:59] <AryehGregor> (there were other times that fsck caught something, but it was some trivial thing that it easily fixed)
# [16:59] <Philip`> whereas an individual can change their filesystem regardless of what anyone else does
# [17:00] <zewt> the problem with python 3 is it doesn't have a sane upgrade path, so it essentially fragments the language
# [17:00] <AryehGregor> zewt, or it could be "we know hardware sometimes goes bad or lies regardless of how good our code is, and we aren't overconfident in our code either".
# [17:01] <zewt> AryehGregor: has never happened to me, but i sure have had servers offline for long periods of time due to an un-asked-for fsck on reboot
# [17:01] <AryehGregor> Yes, that's just stupid configuration.
# [17:02] * Philip` has never encountered the problem of having too much disk space
# [17:02] <AryehGregor> I've had to shrink ext* on occasion. It's really slow and only works offline, but it works.
# [17:03] <zewt> really the only particularly interesting feature of newer filesystems is being able to snapshot (without layering another special block layer underneith), to be able to backup databases directly
# [17:03] <Philip`> If you don't mind really slow, copy the data to an external disk then repartition and reformat then copy back
# [17:03] <zewt> Philip`: well, if you're working with an array, you don't always have that option
# [17:03] <AryehGregor> Philip`, only works if you have enough empty storage.
# [17:04] <zewt> (+ interesting feature to me, of course)
# [17:04] * Philip` wouldn't really want to resize partitions unless he already had a backup copy anyway
# [17:04] <AryehGregor> btrfs has lots of interesting features beyond snapshotting.
# [17:04] <AryehGregor> In-FS RAID looks like it will be really cool.
# [17:04] <AryehGregor> You'll be able to choose different RAID levels for different files.
# [17:05] <AryehGregor> And it doesn't have the same resyncing problems as normal software RAID.
# [17:06] <AryehGregor> So RAID5 for backups (= almost all reads/writes are serial, take up lots of space) but RAID10 for database files (= lots of random read/write), say.
# [17:06] <AryehGregor> Of course, a lot of this stuff is still on the drawing board.
# [17:07] <zewt> more interesting would be intelligent RAID for databases across spinnydisks and SSDs
# [17:08] <AryehGregor> Oh, and you can copy files instantly, like hard-linking except the file is actually logically totally separate. That's very cool.
# [17:08] <zewt> eg. to be able to get safe writes to disk (which many, especially cheaper, SSDs are very bad at), while being able to use the SSD for reads (where it's an order of magnitude improvement for databases in many cases)
# [17:09] <AryehGregor> That's probably best handled at a different layer.
# [17:09] <AryehGregor> Somewhere below the filesystem layer.
# [17:09] <AryehGregor> There's work in that department in Linux, IIRC including something funded by Facebook.
# [17:10] <zewt> it can be done at the FS layer logically: report writes as synced to disk as soon as they've hit at least one disk, without waiting for it to be confirmed to all
# [17:10] <zewt> (whether you actually want that or not is a matter of policy)
# [17:11] <AryehGregor> md already can do that, with the --write-behind= option. Although of course that's inflexible.
# [17:11] <AryehGregor> In particular, the SSD has to be the same size as the other device, which mostly defeats the point.
# [17:11] <zewt> of course, different disk sizes and other things mean it's far from trivial
# [17:12] <AryehGregor> The better use for SSDs is probably to use them as a read cache only, not a write buffer. Cheaper SSDs are very slow to write to randomly anyway.
# [17:12] <AryehGregor> Plus, writes can be buffered a lot already in memory or in battery-backed RAID controllers or such.
# [17:13] <AryehGregor> Whereas you can't buffer reads, so those are the things you really want to get rid of as much as possible (admittedly depending on workload).
# [18:23] <timeless> SJC and DCA seem to be about the same for North
# [18:24] <timeless> I've never lived much further south than that, but..
# [18:24] <timeless> I've really only lived North for the last bit of my life :)
# [18:25] <timeless> and YYZ is really much further south tha HEL :)
# [18:26] <jgraham> You can put a bound on where Ms2ger lives from the fact that he experiences post 10pm local time sunsets but they are currently not post 10pm
# [19:02] <eighty4> timeless: no, I mean coffeescript. But it wasn't my quote
# [19:03] <AryehGregor> TabAtkins, you have to say that it's a great circle, because latitude lines aren't great circles (except the equator). So a degree of latitude will generally be less distance than a degree of longitude, up to 100% less.
# [19:03] <AryehGregor> . . . I mean the other way around.
# [19:03] <AryehGregor> A degree of longitude at our latitude is about 70% as long as a degree of latitude.
# [19:04] <Ms2ger> In Texas, all circles are great circles.
# [19:10] <TabAtkins> AryehGregor: No, you had it right the first time.
# [19:10] <TabAtkins> latitudinal degrees are smaller than longitudinal over most of the world
# [19:10] <AryehGregor> No, because moving a degree of latitude is moving north or south, and that's always along a great circle (one that contains the north and south poles).
# [19:13] <AryehGregor> (a 4-torus, strictly speaking, would be a four-dimensional manifold in its own right, not two-dimensional)
# [19:13] <TabAtkins> My geometry terminology is all kinds of inconsistent.
# [19:14] <TabAtkins> So yeah, a 2-torus embedded in 4d so it's flat in 3d
# [19:14] <AryehGregor> Well, to be fair, nobody talks about these things precisely unless you take graduate-level differential geometry or something.
# [19:14] <Ms2ger> (* Recognizable by the use of the word "geodetes")
# [19:16] * Philip` was thinking more of a discworld (where you just need X and Y coordinates relative to some arbitrary axis) or ringworld (where you need one angular and one linear coordinate)
# [19:17] <Philip`> (I suppose ringworlds are inconvenient for drawing maps though, since you need an incredibly long thin piece of paper)
# [19:17] <TabAtkins> Are we talking a circular discworld? Polar coords are still best there. A square discworld can just use x/y, but you might as well get exotic and go for the torus then.
# [19:17] <Ms2ger> What's next, want it on top of an elephant too?
# [19:18] <Philip`> Ms2ger: Don't be silly - it'd fall off its back
# [19:18] <Philip`> You need at least three elephants for stability
# [19:18] <TabAtkins> Philip`: A ringworld doesn't need to be enormous unless you're trying to build it in physical reality.
# [19:19] <Philip`> TabAtkins: But polar coordinates make it hard to work out distances between points, which is probably the main value in having coordinates at all
# [19:19] <TabAtkins> Philip`: I know, but that doesn't change the fact that polar coords are the most natural for a circular disk.
# [19:22] <AryehGregor> Philip`, not if it's a flat torus. Then it's just a rectangle with opposite edges identified, so it can be whatever proportions you like.
# [19:22] <AryehGregor> It can be a square, no problem.
# [19:22] <TabAtkins> Rama's a perfectly fine ringworld as well.
# [19:22] <Philip`> If it was somewhere in the middle it'd look silly
# [19:22] <AryehGregor> Also, then you can use (x, y) coordinates and calculate using Euclidean distance.
# [19:22] <AryehGregor> It's the best of all worlds.
# [19:23] <AryehGregor> (in multiple senses, I guess)
# [19:23] <TabAtkins> AryehGregor: Though, you still have to do a bit of math. If you're near the, um, dateline-equivalent, it's shorter to measure across that.
# [20:35] <timeless> There exists an algorithm which transforms names and can be reversed losslessly. (Yes, yes, you can do it if your algorithm returns the input. You get a gold star.)
# [20:36] <AryehGregor> Is that the one that basically says "you can't assume anything whatsoever about names, up to and including that people actually have one"?
# [20:43] <timeless> 41. If a person has a title such as “Doctor”, they will have at most one such title. (In Germany, the over-educated go by names such as “Prof. Dr. Dr. Hans Schmidt” – a professor with two PhD’s)
# [20:45] <AryehGregor> Actually, most of the rabbis I know go by Mr. unless they're actually rabbis of a congregation or religious instructors or something. Kind of like how most Ph.D.'s don't call themselves "Dr." or "Professor" outside of a professional context, and you get suspicious if they do.
# [20:45] <timeless> like en.wikipedia.org/wiki/Worf
# [20:48] <AryehGregor> As far as I can tell, basically every man who's brought up Lubavitch and wants to spend time proselytizing is given the title of rabbi for professional reasons.
# [20:48] <karlcow> I wonder if HTML would have been developer during the revolution if it would have been without <head>
# [20:51] <karlcow> I guess in the list of names which can't be written are all the names of spoken only languages
# [20:53] <timeless> > 43 – A person name in country-1 is the same as his name in country-2. In Holland my name is Gert; In Denmark it is Gerardus; In Germany they refuse to try pronounce my name
# [20:56] <Philip`> It's much easier to assign everyone a number and refer to them by that
# [20:56] <karlcow> my Japanese Hanko for papers is not the name on my passport
# [20:56] <timeless> Philip`: i'm actually supposed to be thinking about a Contacts api
# [21:02] * timeless will have to get zewt to look at some APIs at some point to see if they work
# [21:02] <zewt> i think it's some usps api they use to verify shipping addresses ... but usps is wrong
# [21:02] <timeless> presumably USPS will deliver to the (its?!) PO Box
# [21:02] <zewt> my street address, with the PO box in the ZIP code, is valid for shipping to me (it's what i use since it works for usps + fedex + ups)
# [21:03] <zewt> but i think the usps api only accepts the actual po box, and t-mobile's thing screamed WE DON'T SHIP TO PO BOXES
# [21:08] <timeless> > Barbon did this despite long-established restrictions on new buildings associated with various Acts of Parliament and royal declarations in the late 16th century: he often simply disregarded legal and local objections, demolished existing buildings without permission and rebuilt speculatively in search of a quick profit.
# [21:08] <timeless> great guy, clearly a pragmatist :)
# [21:17] <jgraham> Philip`: Move to Sweden, they have the "everyone should be referred to by number" thing sorted
# [21:17] <zewt> google contacts has a "name" and "transliterated name" thing, though that's not really enough
# [21:17] <Ms2ger> jgraham, how hard is it to get a number?
# [21:17] <timeless> zewt: people complain that google's outgoing behavior is problematic
# [21:18] <timeless> but i think that's more because it tries to serialize to forms that have first/last or similar
# [21:18] <zewt> eg. if i have a contact with a japanese name and transcription and a japanese person has the same, it's hard for a UI to know whether the user wants a particular contact natively or transliterated
# [21:18] <timeless> what's amusing is that i thought it was a numbers only systems
# [21:18] <zewt> still, it's a small step forward from contacts that don't have it at all
# [21:18] <gsnedders> Ms2ger: You have to have some address for some time, though not always that long.
# [21:19] <timeless> it turns out that there's a <dash> between two numbers which isn't always a dash
# [21:19] <timeless> and is thus a significant character in the `number`
# [21:19] <AryehGregor> gsnedders, what you're missing is that for most purposes, names need be only locally unique. Your friends would just call you '497. It would be quite practical.
# [21:19] <timeless> yeah, for contacts, your set is generally locally unique
# [21:20] <timeless> and when you give out a card to someone, you probably only ever give out a subset of the card
# [21:20] <timeless> imagine the card for your s/o has `likes pink roses`
# [21:20] <gsnedders> AryehGregor: You don't need the checksum, really :)
# [21:20] <timeless> you wouldn't give out that part of the card when you give it to someone
# [21:20] <Ms2ger> Calling you 1992', OTOH, would not work all that well
# [21:20] <AryehGregor> No, that would be a problem.
# [21:20] <gsnedders> I seem to be mistaken, the official form is 920480-4497
# [21:27] <jgraham> (I believe it is the case that under EU law it muct be possible to use any EU passport in a circumstance where a passport from a single member state is valid)
# [21:28] <Ms2ger> Time for a single EU passport, with a single database to track anything any government knows about you? :)
# [21:28] <gsnedders> jgraham: It's more general than that: you're not allowed to discriminate against citizens of other member states compared with your own
# [21:29] <gsnedders> (There are a few exemptions for a limited time upon new member states joining)
# [21:29] <erlehmann> AFAIK, in germany, the constitutional court banned globally unique numbers. now we have a tax id. haha.
# [21:30] <jgraham> gsnedders: Well it's not really clear to what extent that is actually true e.g. they are supposedly introducing some laws here to make it harder to be a teacher if you don't have a Swedish qualification
# [21:30] <erlehmann> (globally unique numbers for tracking purposes, i mean)
# [21:30] <jgraham> Which seems pretty suspect to me
# [21:30] <jgraham> But I guess they could claim that anyone could get such a qualification, or something
# [21:30] <timeless> erlehmann: sounds like the us
# [21:31] <gsnedders> jgraham: Anyone can get the qualification is the theory there.
# [21:31] <timeless> jgraham: iirc Belgium made it pretty much impossible to teach French if you weren't a native French speaker
# [21:31] <timeless> (or was that if you weren't a Native and a French speaker?)
# [21:32] <gsnedders> jgraham: It is the reason why Scottish universities don't charge tuition fees for anyone from the EU except for the rest of the UK, due to that law and its interaction with the Scotland Act, which forbids the Scottish Goverment to pass an act that would pay for the rest of the UK.
# [21:33] <erlehmann> timeless, germany has a quite strong “data protection” culture and laws. fun fact: since the people responsible for data protection and freedom of information are often the same, the latter is lacking considerably.
# [21:38] <AryehGregor> Yeah, Germans seem to be extremely concerned with privacy.
# [21:38] <TabAtkins> We'd call them the "privacy czar" in America.
# [21:38] <timeless> > Israeli law requires every permanent resident above the age of 16, whether a citizen or not, to carry an identification card called te'udat zehut (Hebrew: תעודת זהות) in Hebrew or biţāqat huwīya (بطاقة هوية) in Arabic.
# [21:38] <erlehmann> yeah, those hollerith machines seem to have hit a nerve there.
# [21:38] <AryehGregor> I'm not concerned about privacy at all, so I'll be a good fit for Israel. :)
# [21:39] <AryehGregor> (not that Israelis aren't concerned with privacy, but they're pragmatic . . .)
# [21:39] <AryehGregor> timeless, I've heard it claimed tourists are supposed to carry photo ID too, although no one ever officially told me that when I was a tourist.
# [21:40] * timeless unfortunately has a hard time holding the meaning of `dati` and some other one
# [21:40] <erlehmann> TabAtkins, even if a company is required to have one? i thought “XXX czar” is reserved for federal overlords?
# [21:40] * timeless uses haredi in place since it has a meaning
# [21:40] <AryehGregor> I'm sure there's a special exception for religious observance.
# [21:41] <TabAtkins> erlehmann: Oh, didn't realize that was a company thing.
# [21:41] <AryehGregor> Dati just means religious.
# [21:41] <timeless> > Criminal offence carries a 5,000 Old Israeli shekel fine for not carrying an identity card or for misuse of the document (in 1983 prices, which equal about 1,400 NIS today).
# [21:41] <AryehGregor> erlehmann, "czar" is a word that newspapers use in headlines to save space.
# [21:41] <timeless> yeah, i can't translate `dati [religious]` into my conceptual map
# [21:41] <timeless> if i go to services friday night, saturday morning, saturday night, sunday morning, and tonight
# [21:42] <erlehmann> TabAtkins, it is both. as i understand it companies have to have people responsible for this. local and federal authorities have to have to.
# [21:42] <TabAtkins> I'd say it makes you religious.
# [21:42] <timeless> ok, and that strictly translates to dati?
# [21:43] <timeless> there's the place i want to go, the place i'd like to go later, and a place i was asked to go to ensure they have a minyan so people can say kaddish
# [21:43] <timeless> (and i haven't unpacked enough boxes to find my canvas shoes)
# [21:43] <AryehGregor> "Chiloni" means you don't believe in or care about religion at all. "Masorti" means you're sympathetic to religion and observe some practices at least when they suit you, maybe including keeping kosher and praying, but aren't fully observant (e.g., don't observe the Sabbath or conjugal laws or something). "Dati"/"charedi" means you're religious, you theoretically try to keep all of halacha. The difference between "dati" and "charedi" is so
# [21:43] <AryehGregor> cial: datim are a part of the secular world, charedim aren't.
# [21:44] * timeless should be able to memorize that
# [21:44] <erlehmann> TabAtkins, i believe that clubs and companies that have at least 10 people who work with critical data need a person responsible for data protection.
# [21:44] <timeless> i'm masorti, and i know that, but i haven't yet learned that it actually fits into that map
# [21:44] <AryehGregor> So for something like carrying on Shabbat, there's no real difference between charedi and dati. Anyone who would knowingly carry on Shabbat without an eruv is almost surely masorti.
# [21:45] <AryehGregor> "Chiloni" is from "chol", "secular". "Masorti" is from "mesorah", "tradition". "Dati" is from "dat", "religion". "Charedi" is from "charad", "to tremble".
# [21:45] <erlehmann> timeless, read the article. i am no lawyer, but i think it is pretty interesting.
# [21:46] <erlehmann> even privacy-unrelated stuff is discussed like this in german culture, so it sometimes gets somewhat silly. google street view is pixelated at some places because residents felt uneasy.
# [21:46] <timeless> AryehGregor: yeah... i should be able to hold that
# [21:46] <timeless> but the problem is that i learned these concepts as English-phonetic-sounds long after i stopped using Hebrew for general things
# [21:47] <erlehmann> AryehGregor, not necessarily. curtailing specific powers certainly prevents abuse.
# [21:47] <timeless> and thus my mapping for them is poor
# [21:47] <timeless> whereas i actually do know the hebrew words...
# [21:47] <AryehGregor> Dati vs. charedi is like Modern Orthodox vs. yeshivish/ultra-Orthodox.
# [21:48] <timeless> yeah, that last one is also a problem
# [21:48] <erlehmann> fun fact: a friend of mine, plomlompom, is writing a (post-)privacy book. i'm curious if it will cause a shitstorm in privacy-related discussions when it comes out in one or two months. ;)
# [21:48] <timeless> since i never really got an introduction/explanation of what modern-orthodox is
# [21:48] <timeless> so i have a hard time drawing/understanding that line
# [21:48] <timeless> i can easily understand reform/conservative/orthodox
# [21:48] <timeless> because i grew up with those concepts
# [21:49] <timeless> erlehmann: presumably in German?
# [21:49] <Philip`> People who are concerned about their privacy on Google Street View should print out life-sized cardboard copies of themselves and leave them all over the town, so that nobody can tell where the real them is
# [21:50] <Philip`> It's pretty much impossible to hide information nowadays, but you can overload people with bogus data so they can't find the real information
# [21:50] <erlehmann> timeless, indeed. but it also includes a quite interesting short history of privacy, so it might be usable for other cultures as well.
# [21:50] <erlehmann> title is „prima leben ohne privatsphäre“
# [21:53] <erlehmann> AryehGregor, i think for me the “germans are crazy” thingy can be exemplified by the fact that in sweden, Stuff Your Gov Does™ is public by default until classified otherwise – whereas in germany, it is not until classified otherwise. or so i heard.
# [21:53] <AryehGregor> Germany has to have *some* kind of classified information, otherwise they couldn't (for instance) have diplomatic negotiations with other countries.
# [21:54] <TabAtkins> erlehmann didn't say Sweden didn't have classified information, just that it's not the default.
# [21:54] <AryehGregor> He said Germany didn't, unless I misread him.
# [21:55] <TabAtkins> You misread. He said it's secret until classic otherwise.
# [21:56] <AryehGregor> No, I was just confused by context.
# [21:56] <AryehGregor> But seriously, someone once sued Wikimedia Deutschland because the English Wikipedia published the names of two convicted German murderers. Apparently under German law you aren't allowed to publicize the fact that someone committed murder after they've served their sentence.
# [21:57] <AryehGregor> It's a violation of their privacy.
# [21:57] <erlehmann> yeah, i think the reasoning is because it damages their social standing more than intended by the law.
# [21:58] <AryehGregor> (they got the judge to make Wikimedia Deutschland take down its wikipedia.de redirect, so it put up a page saying "We had to take down this redirect because of court ruling X. Please donate!", so the plaintiffs changed their mind and decided they were okay with the redirect after all)
# [21:58] <AryehGregor> (this is why no Wikimedia data is hosted anywhere outside the United States, and the only non-data servers that are hosted outside the United States are in the Netherlands)
# [21:59] * Quits: enyo (~mtl@modemcable098.129-202-24.mc.videotron.ca) (Quit: Lost terminal)
# [21:59] <Hixie> the US has a "default-free" mode too, as far as i can tell, but that doesn't stop the US classifying everything under the sun
# [21:59] <AryehGregor> Nowhere in the world has free-speech guarantees close to the U.S.'s, that's for sure.
# [22:00] <AryehGregor> Hixie, actually I don't think that much is actually classified gratuitously. It's a pain to request documents, but if you really want to pursue it you can take it to court and everything, AFAIK.
# [22:00] <AryehGregor> Of course, there are still a bunch of exceptions, but you can get a fair hearing by a judge and everything, if you want.
# [22:00] <Hixie> yeah maybe the "default-free" is really "default-hidden-until-requested"
# [22:00] <AryehGregor> (and are willing to spend the time and money)
# [22:01] <AryehGregor> Right, that's probably a better description.
# [22:01] <TabAtkins> AryehGregor: Money isn't really an issue - FOIA requests are free, I think.
# [22:01] <TabAtkins> Time, definitely, becasue they'll stall sometimes.
# [22:01] <AryehGregor> I mean, nobody's going to say you're allowed to wander in to a government office and browse the files on some random computer.
# [22:01] <AryehGregor> TabAtkins, it's probably free to file them, but if they deny it and you want to challenge it, I'm pretty sure you'll need a lawyer.
# [22:02] <erlehmann> there also was a boxer called “neger-kalle” (literally: “nigger kalle”) who (successfully, i think) sued a newspaper because they had it in the archives and now he says it is racist.
# [22:03] <erlehmann> freedom of information is a different thing than free speech.
# [22:04] <gsnedders> AryehGregor: Most countries have laws about whistleblowers. Most places will still find some technically legal way to screw you over.
# [22:05] <gsnedders> AryehGregor: Depends if it is classified or not. If it's not classified, in most places they theoretically do. Just see the above.
# [22:06] <AryehGregor> Although I don't know, maybe they could convict them but are afraid of the PR backlash.
# [22:06] <erlehmann> Hixie, that also sounds like black magic!
# [22:07] <AryehGregor> Anyway, it would certainly be protected by free speech in some cases. Look at New York Times Co. v. United States, re the Pentagon papers.
# [22:07] <Hixie> AryehGregor: you mean the way that US gov't officials have said assange should be assassinated?
# [22:08] <AryehGregor> Hixie, I said it provided some protection. I don't think anything's going to protect Assange, his leaks were much too indiscriminate.
# [22:08] <Hixie> ah, free speech, but only if you discriminate, ok
# [22:09] <AryehGregor> Well, no one said free speech is unqualified. There are limitations.
# [22:09] <karlcow> I have a name for the patrons but that will not work in that community
# [22:10] <Hixie> i don't see any limitations in the text of the constitution
# [22:10] <Hixie> must be one of those "interpretation" things
# [22:11] <Hixie> i guess assassinations wouldn't be covered, since those wouldn't count as laws?
# [22:11] <TabAtkins> Basically, if your speech is threatening, libelous, was done knowingly to cause harm, or violates copyright.
# [22:11] <AryehGregor> Well, yes, the Constitution wasn't really meant to be interpreted in isolation from the entire tradition of common law that preceded it.
# [22:12] <AryehGregor> TabAtkins, under court precedent, the state can impose any restriction on free speech provided it can show that a) there's a compelling state interest, b) the restriction is narrowly tailored to serve that interest.
# [22:12] <AryehGregor> It's strict scrutiny, that's all.
# [22:12] <AryehGregor> There are certain exceptions where the state isn't even subject to strict scrutiny, based on the traditional application of free speech historically in America, like obscenity.
# [22:13] <gsnedders> AryehGregor: So basically they can argue any leak of confidentical material puts state security at risk?
# [22:13] <Hixie> AryehGregor: you're saying the constitution is more like HTML4 than HTML5? :-P
# [22:13] <AryehGregor> gsnedders, they can argue that. It's up to the judges to decide on a case-by-case basis. I'm pretty sure judges are not going to say that Assange's leaks are fully protected, if it came to a trial.
# [22:14] <zewt> free speech means "free speech unless someone finds it offensive" these days
# [22:14] <AryehGregor> Hixie, sheesh, actual *laws* are more like HTML4 than HTML5. The *Constitution* is like HTML4 with all the precision removed.
# [22:14] <AryehGregor> If you tried to apply the Constitution literally, you'd have all sorts of crazy and totally unintended consequences.
# [22:14] * jgraham wonders if Hixie was responsible for the "server/slave" terminology in the Opera test systems
# [22:15] <AryehGregor> Example argument: since the right to keep and bear arms shall not be infringed, all American citizens have the right to own and operate nuclear weapons.
# [22:16] <jgraham> I wouldn't know a stargate reference if it ate my face
# [22:16] <gsnedders> Hixie: (Having never seen Stargate I've probably not noticed them)
# [22:17] <jgraham> Now we have portal references...
# [22:17] <TabAtkins> AryehGregor: Are you kidding? Freedom of religion is the only thing that stops the dominionist christains from taking over.
# [22:17] <zewt> let's use one etching-away-of-constitutional-freedoms to attack another constitutional freedom, just for irony's sake
# [22:17] <AryehGregor> Also: since the Thirteenth Amendment bans involuntary servitude, the government is not allowed to make anyone do anything at all, such as go to school.
# [22:17] <Hixie> AryehGregor: yeah, the lack of a limitation of rights to the extent that they infringe on other people's even more fundamental rights is problematic
# [22:17] <AryehGregor> TabAtkins, you could remove it and replace it with something else, like mandating a completely secular state (which America never was).
# [22:18] <Hixie> AryehGregor: (i.e. the constitution doesn't say how to resolve the case of two conflicting religions)
# [22:18] <Hixie> gsnedders, jgraham: the machine names were all named after stargate characters originally
# [22:18] <Hixie> gsnedders, jgraham: for spartan, that is
# [22:18] <AryehGregor> Hixie, but this stuff was all *extensively* discussed by philosophers and political thinkers around the time the Constitution was written. There was a common understanding of what a lot of this stuff meant.
# [22:20] <AryehGregor> TabAtkins, is not. It says that Congress shall pass no law restricting the free exercise of religion. The government could be explicitly Christian as long as it doesn't stop people from practicing other religions. In fact, that was basically the status quo until sometime in the mid to late 19th century.
# [22:20] <AryehGregor> Plus, the First Amendment never applied to the states at all until after the Civil War. There were states that said only Christians could vote.
# [22:21] <gsnedders> Hixie: Allan ended up using Pokémon when far more slaves were acquired, FWIW. But even those are massively outnumbered.
# [22:21] <TabAtkins> AryehGregor: No, it must "make no law respecting an establishment of religion". An explicitly religious government would, by necessity, have laws respecting the state religion.
# [22:22] <AryehGregor> TabAtkins, well, no, the federal government wasn't allowed to have a state religion along the lines of the Anglican Church. It still didn't have to be secular in the sense of being neutral with respect to religion, or certainly not neutral with respect to atheism. When were atheists first allowed to testify in Court in the US?
# [22:23] <gsnedders> jgraham: That's unfair. There are only 12.
# [22:23] <TabAtkins> There were no federal laws preventing atheists from testifying.
# [22:27] <dbaron> In the period when the supreme court was most aggressive about enforcing rights, they didn't particularly care about enforcing freedom of religion.
# [22:27] <dbaron> In fact, Earl Warren (Chief Justice) previously won a campaign for Governor of California by basically attacking his opponent for athiesm.
# [22:28] <jgraham> karlcow: They made a fundamental mistake with the US. What you really want in a country is an official religion that is rather hands off. The church of England is practically perfect in this regard. Only the queen is expected to be religious and the rest of us can happily ignore it
# [22:29] <jgraham> By not forcing people to construct their identities by making positive choices around their religion it is much easier to slip into casual atheism
# [22:29] <zewt> can we just opt for an athiest government
# [22:29] <dbaron> karlcow, btw, many of the things that include religion were done in the early 1950's.
# [22:29] * AryehGregor will soon be moving to a country where ~20% of the parliament is pro-theocracy
# [22:29] <karlcow> dbaron: ah interesting. I didn't know
# [22:30] <dbaron> karlcow, things like "In God We Trust" on the money, etc., from the early 1950's.
# [22:30] <Hixie> sicking: re bug 538142 -- i would take one implementation as an indication that the direction of implementations was that way, and change the spec accordingly.
# [22:30] * AryehGregor cannot easily find any sources about atheists' eligibility for testifying in court
# [22:30] <Hixie> sicking: my goal is to have the spec slightly ahead of implementations, and going in the same direction.
# [22:31] <AryehGregor> dbaron, on the other hand, things like banning prayer in public schools or banning public government displays that include religion are even more recent, and AFAIK have no precedent before the last few decades.
# [22:31] <sicking> Hixie: well, given that currently 4 browsers have said they are not interested, i wouldn't take one chainging their mind as a sign that the 3 others would
# [22:31] <dbaron> AryehGregor, Have we actually done the latter?
# [22:31] <AryehGregor> "In God We Trust" appeared on US coins since 1864, by the way, according to Wikipedia.
# [22:31] <TabAtkins> AryehGregor: Prayer has never been banned in schools. Prayer at public events, on the other hand, has.
# [22:32] <gsnedders> FWIW, in Scotland, the Church of Scotland is by law a separate entity from the state entirely, and neither can interfere with each other. This does actually come up every so often, and there is even more separation than in England.
# [22:32] <AryehGregor> TabAtkins, I mean school-organized prayer, like in the classroom.
# [22:32] <TabAtkins> AryehGregor: Yes, that's certainly been banned widely in the last few decades.
# [22:32] <AryehGregor> My father went to public school and they still had Bible readings.
# [22:32] <dbaron> AryehGregor, ah, ok, but it wasn't required on coins until 1956
# [22:32] <TabAtkins> Under the (reasonable) doctrine that public school is an accessory of the government, and must obey the same restrictions.
# [22:33] <gsnedders> (For the curious, look up the Church of Scotland Act)
# [22:33] <Hixie> sicking: typically when one browser does something (especially mozilla) it tends to influence the others, so it's not that simple. :-)
# [22:33] <Hixie> sicking: anyway this seems largely academic since you don't want to do it anyway :-)
# [22:34] <sicking> Hixie: asking also works sometimes...
# [22:34] <AryehGregor> TabAtkins, there were no bans on any government-organized prayer or Bible readings before the last few decades, AFAIK. Obviously, students who objected could always opt out.
# [22:34] * AryehGregor has done *nothing* useful today :(
# [22:44] <jgraham> (because it seems to offer a convenient get out clause from having the humility and introspection to admit when you have made a horrendous mistake)
# [22:45] <jgraham> ("God will be my judge in Iraq" and all that)
# [22:53] <sicking> Hixie: safari disabled 3rd party by default
# [22:53] <sicking> Hixie: we're actually trying to fix the privacy situation, that's why we're avoiding digging deeper
# [22:54] <Hixie> you could drop cookies entirely, and it wouldn't help at all
# [22:54] <gsnedders> sicking: We disabled 3rd party by default in 10.50, but it broke major sites (mainly in, e.g., Russia, where Safari has almost no marketshare)
# [22:54] <Hixie> there's far too many bits of unique data per user to ever stop fingerprinting
# [22:58] <Hixie> (i mean in particular in this shared worker case, instead of a shared worker to correlate sessions, you could just do an xhr to a remote host -- the IP address plus the HTTP headers of the request would be more than sufficient to associate the requests as being from the same session)
# [22:58] <Hixie> so in other news, apparently mozilla prevents ws:// websockets from being opened from https:// pages. Which I guess makes sense if the socket is to something sensitive used by the page
# [22:59] <Hixie> doesn't really make sense if the target websocket is some third-party trivial thing like a clock service or a stock ticker or something
# [22:59] <Hixie> should we put that restriction in the spec?
# [23:02] <Philip`> Pages might treat the output of the service as HTML and insert it directly into their content, which would be an XSS hole if the output could be tampered with by an attacker, and HTTPS is meant to prevent such tampering
# [23:03] <Hixie> oh it's certainly possible to shoot yourself in the foot if we allow https:// to ws:// connections
# [23:14] <smaug____> there should be some status page for "all" the browser/spec developers
# [23:14] <Hixie> Philip`: if you could contribute your insight on the canvas drawimage thread on whatwg that would be awesome
# [23:15] <Hixie> Philip`: especially if you can convincingly argue for not changing the spec :-)
# [23:15] <TabAtkins> If I had some engineers interested in putting together a "report a problem to the developer API" that could be invoked and then would give the user a browser-generated dialog allowing them to choose to submit a comment, machine details, and screenshots to the page author, where would be best to have them propose this?
# [23:26] * Quits: kor (~kor@ip146-53-210-87.adsl2.static.versatel.nl) (Quit: kor)
# [23:27] <sicking> Hixie: what do you mean by "Web Sockets by definition are same-origin with whatever script created them"? I would say that it's cross-origin by definition since it always connects to a different scheme
# [23:34] <Hixie> sicking: well i guess it depends what you mean by "origin"
# [23:35] <Hixie> sicking: on the web you can only access same-origin resources. The way I look at it, CORS and WebSockets get around this by allowing servers to opt-in to being treated as same-origin in specific cases.
# [23:36] <sicking> Hixie: that's not how i see CORS
# [23:36] <sicking> Hixie: i see it as a way for a server to say "you can read this data even though you are from a different origin"
# [23:36] <Hixie> what's the origin of an <img> obtained using CORS?
# [23:36] <Hixie> of the image data of an <img>, i mean
# [23:37] <Hixie> it's not cross-origin; if it was, it would still taint canvas
# [23:37] <sicking> Hixie: it's the origin of the server it was loaded from