|
View previous topic :: View next topic
|
| Author |
Message |
Won
Joined: 21 Sep 2005
Posts: 506
Location: New York
|
| |
 Posted: Fri Aug 15, 2008 4:28 am Post subject: |
 |
|
Besides the Word/Framemaker revelation, what else made you think badly of him? I didn't really have the patience to slog through the hour. Maybe you should put in in your non-fiction review.
You have to respect the commitment to write the CSS thesis using CSS...although it is much cooler when the technology is good like SVT.
I wonder if LaTeX had hyperlinks whether HTML/CSS would ever have happened. |
|
| Back to top |
|
 |
casey
Site Admin
Joined: 18 Dec 2004
Posts: 1768
Location: Seattle
|
| |
Posted: Fri Aug 15, 2008 6:40 am Post subject: |
|
|
| Won wrote: | | Besides the Word/Framemaker revelation, what else made you think badly of him? |
The entire lecture, people from Google kept asking fairly important questions, like "what about widow control?" and stuff. And usually, they did it in the context of actually pointing out a typesetting _problem_ on the guy's slides that were meant to be demos.
And every damn time, both the dudes were like, "yeah that's not really the focus... we're just trying to handle 80% of the documents". I don't really know what he meant by that. He kept saying 80% like somehow, looking like shit because you don't do good typesetting is "OK" 80% of the time. As far as I'm concerned, it's OK 0% of the time. You either look like shit or you don't. It's not like sometimes you don't care if you look like shit, because otherwise, why were you using a style sheet at all?? Just put raw text in a <body> tag! The whole point of style sheets is to lay out documents in an attractive, easy-to-read fashion.
So it just seemed to me that not only did they not know anything about typesetting, but they also didn't care, and somehow had convinced themselves that nobody else cared either (well, 80% of the people didn't care).
- Casey |
|
| Back to top |
|
|
ryg
Joined: 31 May 2007
Posts: 276
Location: Kirkland, WA
|
| |
Posted: Fri Aug 15, 2008 3:51 pm Post subject: |
|
|
| casey wrote: | And every damn time, both the dudes were like, "yeah that's not really the focus... we're just trying to handle 80% of the documents". I don't really know what he meant by that. He kept saying 80% like somehow, looking like shit because you don't do good typesetting is "OK" 80% of the time. As far as I'm concerned, it's OK 0% of the time. You either look like shit or you don't. It's not like sometimes you don't care if you look like shit, because otherwise, why were you using a style sheet at all?? Just put raw text in a <body> tag! The whole point of style sheets is to lay out documents in an attractive, easy-to-read fashion.
So it just seemed to me that not only did they not know anything about typesetting, but they also didn't care, and somehow had convinced themselves that nobody else cared either (well, 80% of the people didn't care). |
Didn't watch all of it, but the first 25 minutes or so, and what really put me off was how they really didn't seem to have an idea of what problem they were even trying to solve, or what they intended to do better than existing solutions. They really have no clue at all!
Case in point (pretty close to the beginning): First, the CSS guy has a laugh about the Guardian since they do a "newsflash" one-page PDF thingy for their website every hour, and how it's apparently done in InDesign - I quote: "I think InDesign is gonna be... you know, these people who worked with lead in the old days? And then computers came, and they were sort of sitting in the corner, still doing lead for a while... I think InDesign is gonna be like that, you're gonna have that one InDesign guy in the corner, and they're probably gonna have a union...". And then for the rest of the talk he's standing there and admitting that yeah, this or that may not really be what you want, but you can still hand-tweak it and besides, they're "only targeting 80% of the documents". Well, guess what... a newspaper, made by people with actual knowledge of typography, that's probably not gonna be in those 80%. And the InDesign guy, he's not paid for copy&pasting the headlines or for "doing all that clicking". He's not paid for setting up the font on every item, since gosh, DTP apps actually have style sheets too, and have had them for decades (who would've thought!). He's the guy whose job is to make sure that it actually looks good, because that's the one thing apps can't do automatically (so far), and CSS (or any kind of style sheet, for that matter) is really completely orthogonal to that issue, and they apparently don't even realize that.
Then people ask them about line-breaking issues, and the Prince guy admits that "yeah, LaTeX does this better, but that's an algorithmic issue and we're working on it". Seriously, WTF?! The Knuth-Plass line breaking algorithm is over 25 years old and well documented. The TeX source code is open, and in book form to boot, with a very detailed description about the inner workings of the algorithm and implementation issues. If you're marketing your program as "a great way of getting web content onto paper", what exactly is your excuse for doing notably worse than an open-source program that's from the early 80s? Especially if you're claiming that existing DTP apps are obsolete, since guess what, by now all of them use that algorithm or improvements of it (took them long enough though).
Now, not having the ambition to be the ultimate solution to a problem is a good thing in general, since at some point the cost outweighs the gain. But they really have that completely backwards. Normally, you'd try to solve the hard problems, and settle with something that is good enough in most cases - to keep complexity to a reasonable level. And then you add a few tuning knobs for the cases where it doesn't work automatically. But what they do is only ever look at 80% of the problems (the easy ones), make a half-assed solution that's extremely complex and full of exceptions for no good reason, and then add 100 tuning knobs because it never does what you want anyway. What's wrong with those people? |
|
| Back to top |
|
|
casey
Site Admin
Joined: 18 Dec 2004
Posts: 1768
Location: Seattle
|
| |
Posted: Fri Aug 15, 2008 7:28 pm Post subject: |
|
|
| ryg wrote: | | What's wrong with those people? |
And what's wrong with us that we're letting this happen?
I think about that a lot lately. The web is a complete disaster, and I'm getting the sinking feeling it's because no good programmers are spending their time on it. So you end up with these idiots being the only people who actually ever proposed a style system for the web, then it gets standardized and adopted, and then the whole world is stuck with it.
- Casey |
|
| Back to top |
|
|
ryg
Joined: 31 May 2007
Posts: 276
Location: Kirkland, WA
|
| |
Posted: Fri Aug 15, 2008 10:13 pm Post subject: |
|
|
| Casey wrote: | And what's wrong with us that we're letting this happen?
I think about that a lot lately. The web is a complete disaster, and I'm getting the sinking feeling it's because no good programmers are spending their time on it. So you end up with these idiots being the only people who actually ever proposed a style system for the web, then it gets standardized and adopted, and then the whole world is stuck with it. |
Well, here's my stance on it: For most problems that I rant about, I know enough about the issue to articulate what's wrong with what I have, but not enough to propose a solution that's actually better for a wide range of problems. Because it's easy to improve on individual things, but it has to form a coherent whole, and that's really the hard part. Same thing with everything in coding, really - the problem is not displaying a Teapot with Awesome Mapping, it's integrating the Awesome Mapping with whatever else you have. And so I don't want to be the guy who publishes his paper on Awesome Mapping with some sample code, because I know that's not going to solve the actual problem.
But every once in a while there's a specific problem that's either interesting or annoying or pressing enough for me to really get into it, and that's usually when my best work happens. That's how kkrunchy happened (the depacker for the last LZ-based versions of kkrunchy is still among the best individual pieces of code I've written; the compressed data is typically ~10% smaller than ZIP for what I was interested in, and the size-optimized depacker was less than 350 bytes). That's why I wrote kkapture (specifically, because the videos for the 2004 scene.org Awards at Breakpoint 2005 were captured with FRAPS and pretty jerky and it annoyed the hell out of me). And the same goes for the DXT compression code and a bunch of other stuff I never published for various reasons. And I can't just sit down and say "okay, now I'm gonna figure out a clever/good/whatever solution for this arbitrary problem", because it's not enough to just sit down and work; you need to care for the whole thing.
I think that's the main problem about PHP, for example. That's not something where someone had a good idea and fleshed the idea out and then wrote the code - it probably started as a slightly better version of server-side includes and only ever acquired new features when something was painful enough for someone to actually change the language. It practically oozes that air of "nobody ever cared enough about me to write me properly". Everybody who writes PHP code just tries to storm in, write the minimum amount of code necessary, then run out and take a shower ASAP.
And the last thing the web world needs is another system that's just barely acceptable for what it was meant to do, written by people who'd rather do something else. There's way too much of that already.
What I intend to do is continue solving these small problems that are still relevant (some more, some less so, admittedly), and do it well, then make the code available if I can. It's not huge steps towards a brave new world, but it's small steps towards a slightly better tomorrow, and I can live with that.  |
|
| Back to top |
|
|
casey
Site Admin
Joined: 18 Dec 2004
Posts: 1768
Location: Seattle
|
| |
Posted: Sat Aug 16, 2008 4:09 am Post subject: |
|
|
| ryg wrote: | | Well, here's my stance on it: For most problems that I rant about, I know enough about the issue to articulate what's wrong with what I have, but not enough to propose a solution that's actually better for a wide range of problems. |
Yes but that's actually what I mean. Obviously I don't know anything about typesetting either. But if either of us decided to take the time to learn about it, we could probably do a really good spec. But we don't. Hence, there just aren't a lot of good programmers around who are taking the time to figure out what needs to be done in these areas of web development, and thus it is not surprising when things turn out well beyond crappy.
- Casey |
|
| Back to top |
|
|
ryg
Joined: 31 May 2007
Posts: 276
Location: Kirkland, WA
|
| |
Posted: Sat Aug 16, 2008 11:59 pm Post subject: |
|
|
Warning: Another loong post/brain-dump. Just happened, sorry .
Yeah, but my point was this: It's not like I've been slacking off during that time. In the time it would take me to really get into one area that I really don't have much of a clue about, I did solve a few interesting and small but nontrivial problems that are way closer to my area of expertise, and did so in a way that also benefited other people. Had I done other things or advertised my stuff better or whatever, I might have made a bigger impact in the sense of having contributed to something more important, but that's as much luck as anything else; for example, the early Linux adopters/kernel hackers probably all were a notable positive force for the computing world in retrospect, since they were all part of making Linux what it is now; but at the time, they were just a bunch of people working on a toy project that they thought was cool. In retrospect, it's obvious that HTTP and HTML and CSS were things that should better be done right the first time; but when it was developed, it was just a tiny group of enthusiasts, it wasn't even clear that the WWW had any big future (or the rest of the Internet, for that matter), there was still Gopher around, and really everything was very much ad-hoc. Work on CSS1 (which introduced misfeatures such as the padding-is-not-inside-width thing) started in 1994; I doubt that anyone really cared for its details back then. This was before even HTML 2.0 was a standard, and any sort of control over the visual display of web pages was an improvement over what was there. Of course, in hindsight it's pretty obvious that there were major problems, and that someone with actual practical knowledge of typography and layout should've worked on it. But that's just because now, it's actually being used by people interested in typography and layout; back then, it was literally about being able to change the font size and weight of H1 elements.
Now of course, that was then, and theoretically there would still be time to do something better. But by now, it's pretty obvious to anyone that the chances of getting a replacement through is practically zero; even if it was the best thing ever, it would never survive W3C politics (since there's CSS already, and "it works good enough, and we all implement it already"). That's the general problem with anything Internet-related - you have to get all the details right, long before anyone starts using it, because you just might be successful and your protocol might turn out to be critical infrastructure 5 years from now.
In fact, that's the case even with the "genius-designed" parts of the net; IPv4 is running out of addresses and DNS has serious security problems. IPv6 is worse than IPv4 in a lot of ways (the design by committee clearly shows), but it's been around since 1996, everyone knows they will have to make the switch at some point, and it's still happening at a glacial pace; DNSsec took ages, it's complex, support for it is scarce, and there's lots of politics involved with the certification business. But it's all critical updates for critical infrastructure, and it's still taking ages to push it through. I just don't see anything like that happening for something that "just" makes the lives of everyone better. That's the great problem with the net - by the point something starts being popular and its defects start to show, it's already too late to sensibly replace it with something better.
Of course, the server-side stuff is easier to upgrade; but there, I actually do see things getting better. When I first did some server-side web code, the "state of the art" was Perl CGI scripts. Now, PHP definitely sucks, but it's still so much better than the Perl stuff that it's not even funny. Haven't looked at Ruby on Rails, and I assume it still sucks to work with, but from what I gather it at least has somewhat systematic naming conventions and libraries that are actually designed (not sure how well, as said, haven't looked at it) instead of just having happened at some point. So that field may suck, probably since most of the users are people who really can't code properly (and don't care anyway), but there's still some real evolution visible there. It would probably go a lot faster if there were more good programmers there (plus the right salesman-type guys to make people listen to what the good programmers have to say), but for me, that's no reason to feel bad about not being there. Because if these people cared more about their development processes, and good library design, and about getting better at what they do, they'd figure this out on their own. It'd be one thing if the web-dev world was working on really hard problems and needed all the brains they could get to figure this shit out. But they're not. The majority of them is working on relatively easy problems that've been solved ages ago, and all the difficulty is just there because they're working with shitty tools which are judged by buzzword compliance instead of functionality - by their own choice. I really don't have a lot of pity for that crowd. (End of web-dev rant)
Anyway, going back to my first point: things like CDep, CMirror, TabView, stb_image, stb_vorbis, Sizer or my DXT code, they might end up being important in some way (Seans code in particular sure seems to crop up in a lot of projects I've noticed lately ) or they might not, but they're out there and they're useful to people. It may not be a big contribution, but it's a contribution, and it's making the world better
(if only in a small way), which is more than can be said for most celebrated "innovations" nowadays (XML, anyone?). We might have potential to do more good for the world than we're doing currently, but at least we're doing some good, which is more than most people. |
|
| Back to top |
|
|
sean
Joined: 01 Feb 2005
Posts: 1392
Location: Kirkland WA
|
| |
Posted: Sun Aug 17, 2008 9:18 pm Post subject: |
|
|
I sincerely hope that in 50 or 100 years there will be way more BSD-ish/public domain code throughout computing. Right now, that stuff dominates on the low-level net, I imagine primarily because it was nearly necessary for interoperation to happen. GPL is a step, but I'm unconvinced it's the right step, because it so strongly excludes serious commercial development.
But one of the nice things about PD/BSD/etc. is "no takebacks". The amount of free code available is monotonically increasing. Of course that only helps if the code is good code, or at least usable code, and if people find out about it.
Due to the monotonic increase, obviously there will be "way more" such code in 50 or 100 years, but I mean as a percentage. I hope stuff like the mainstream OSes, dumb little utlities, etc. etc. will be 'open' and such, and that mainstream commercial applications leverage a lot more free/open tools. I hope games are built on core engines that do a lot of the basic technological stuff and are free.
I'm not sure I think this is likely, but it is the course I'd like to see things go, and I'm trying to do my part to help things go in that direction.
Basically, the idea is just the whole information isn't scarce, it's freely recopyable, and our society will be best served by leveraging that. I don't think that makes sense for movies and TV -- there's no way our society is served by taking away copy control from creators and preventing them from making money, because not very much builds on entertainment products, so if you take away copyright you take away any incentive to create on the scale they create. But software builds on software so much that when it comes to software, I think society is served by large, rampant copying of software. The trick is just finding an incentive for creating that software in the first place. In my case, I mostly create software because I need it myself. That's the incentive for creating, and the incentive for sharing is because I can, I think it's good, and it doesn't hurt me financially. I'm not worried about helping my competitors at my own expense. But I can understand not everyone feeling that way. (And this changes a bit now that I work for RAD; I do have to be a little cautious about helping RAD's competitors, or making RAD's customers not need RAD's products, in what I do with code I write.) |
|
| Back to top |
|
|
Won
Joined: 21 Sep 2005
Posts: 506
Location: New York
|
| |
Posted: Sun Aug 17, 2008 10:43 pm Post subject: |
|
|
Re: Web 2.0 tech
Some examples of new "smart" approaches include:
Volta
GWT
App Engine
XML: a retarded innovation
Re: tubes
People generally know of the problems with HTTP and DNS, but BGP also has its issues. Routing protocols tend not to get the same exposure.
Re: open source
There are many folks at work who are committed to supporting open source. I've been told that a good, commercially friendly license is Apache 2.0, but if folks are willing to go as permissive as PD, great! |
|
| Back to top |
|
|
sean
Joined: 01 Feb 2005
Posts: 1392
Location: Kirkland WA
|
| |
Posted: Mon Aug 18, 2008 1:09 am Post subject: |
|
|
Actual quote from the GWT talk:
| Quote: | | Now once you know that, you eliminate an extra hop at runtime, which is great, because as we saw before, that costs 10% slower just to get this abstraction -- that's No Good. |
It's really sad that it takes 30 minutes in that talk to get to 'we compile different .js files for each user agent that's automatically customized to that user agent'. |
|
| Back to top |
|
|
sean
Joined: 01 Feb 2005
Posts: 1392
Location: Kirkland WA
|
| |
Posted: Mon Jan 11, 2010 8:01 am Post subject: |
|
|
A rather late update on the most important element from this podcast...
Zombie Tapioca Lovefest 4000 has an IMDB page.
Also that page links to a trailer. (Fukkin' ads.)
Also it came in 2nd place for "Humor" at HUMP 3. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
FAQ
Search
Usergroups
Log in
Register
