Slashdot Log In
Ars Technica on OSX/Aqua
Posted by
Hemos
on Mon Jan 31, 2000 09:03 AM
from the very-cool-overview dept.
from the very-cool-overview dept.
Gilmoure writes "John Siracusa has written an excellent article on the technology (Quartz) behind Aqua and its possible impact on GUI industry. " The continued evolution of OS/X has been interesting, even simply from the marketing perspectives. John's take is a good one to read if you haven't followed OSX very much.
This discussion has been archived.
No new comments can be posted.
Ars Technica on OSX/Aqua
|
Log In/Create an Account
| Top
| 259 comments
(Spill at 50!) | Index Only
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
No Theme Support (Score:5)
One of the things I hate about the Apple way of doing things is that they think the way to make things simpler is to take away options.
For instance, we are using CIPE-over-PPP-over-ATM to network ourselves over DSL to another network. It works really well, but it eats up about 60 bytes out of each TCP packet. This is no problem for places where MTU path discovery works, but many places block ICMP, and so MTU discovery doesn't work... in any case, the best solution is to set the max MTU size down to 1440 or so.
Under Linux, this is easy... ifconfig eth0 mtu 1440
Under Windows, there is a registry setting. Not the right way to do it in my opinion, but at least it can be done.
There is NO WAY to do this under MacOS that I could find. The only solution was some guy who hacked up the TCP/IP stack and wrote a little control panel where you could change the settings. That's not the way to do things.
Or the mouse. I know that Apple people say that one button is easier than 2 or 3, but my Mac friend told me the other day that one of the things I needed to do for something was option-click-click-and-hold. That's EASIER than right-click or middle-click?
Now, Apple has done some very good things in terms of user interfaces... it's a very uniform user interface. Back in 1984, it was an extremely modern way to do things. But over the last 15 years or so they've fallen behind in the technical arena. No preemptive multitasking (until now, more on that in a minute) is unacceptable. And how do they make up for it? FUD. Steve Jobs said that you couldn't buy a faster computer than a G3. Not only could you get a PII to run faster, but he completely neglected the Alpha, UltraSparc, PA-RISC, etc. These aren't typically home machines, sure, but he was trying to say that the G3 was some sort of Super Computer or something. We see this continued with the silly Army Tank / G4 commercial, which is not so much a testament to the speed of the G4 (Don't get me wrong, it's a nice chip, but it's not beating the Origin 2000 or Enterprise 4500 behind me any time soon) as it is backwards and outdated US laws.
Now we have OS X. I must say that I'm very happy that Apple is getting into the Modern OS Architecture arena. And they certainly chose some good technologies to support. I have high hopes that Apple can come up with a really excellent product... though I still see that they treat seperate partitions as seperate filesystem spaces. D'oh.
But I must say I'm not totally convinced that OS X is something that I want to run in the future. Apple has not been what I'd call a friendly company in the last few years. It used to be that Apple was the good guy and IBM was the bad guy. Now Microsoft is certainly the bad guy... but I'm not convinced that Apple is a good guy. If Steve Jobs and MacOS controlled 80% of the market share, would that really be better than it is now? Think about who controlls the industry and how they deal with specs and such. Is Apple any better? Worse?
I'd say that it'd be worse. You'd have to buy your hardware from Apple. Prices would be inflated without the competition. And your computer would have to be smurf-puke blue. :-)
Not the mouse again... (Score:3)
When I bought my PC, it came with a 2 button mouse, but I wanted a 3 button so I could use the middle button in netscape to open links in a new window. So I bought a 3 button mouse.
When I bought my Mac, it came with a 1 button mouse, but I wanted a 3 button mouse so I could use the right button to bring up contextual menu panels and program the middle button to do other cool tricks. So I bought a 3 button mouse.
OS X: A First Look (Score:4)
http://www.asktog.com/columns/034OSX-FirstLook.
Jazilla.org - the Java Mozilla [sourceforge.net]
Re:what is so revolutionary about this ? (Score:3)
Did you consider actally *shudder* reading the article? The technology is completely different than anything MacOS or any other OS has done before. That's what's so revolutionary.
Re:Anyone know how the filesystem will work? (Score:3)
Well, in a traditional OpenStep/NextStep/MacOSXServer environment, apps are a container directory. For example, to run TextEdit, you actually run TextEdit.app which is a directory. In the directory are icons, the binary, other app data, and dynamic libraries (Open step is really big on delaying as much as possible to runtime, with heavy use of dynamically loadable modules for interface, etc). I assume OSX will do along the same lines, only making the directory wrapper more transparent to the user.
I don't know how classic apps are going to work though.. In Server, the MacOS apps have to be run of an HFS image or disk. I don't know if/how they are going to make these run on UFS. Maybe you'll have to keep a MacOS disk image around to run Classic apps.
Re:Quartz is great! (Score:4)
It currently suffers from the ESR RampUp syndrome - i.e. the necessity to have a minimally working product before the OS scale-up really begins to have an impact.
They are basing it on OpenGL and GGI. Its called 'Berlin'. They've got a page at SourceForge.
Nothing actually usable in terms of replacing X yet though.
Then there's always the GNUStep people... I think they're trying to use 'Display Ghostscript'.
All things considered, I don't think Quartz is a really monumental achievement. After all, they had NeXT's implementation of Display Postscript to work from. Considering how quickly gv was adapted to display PDF files, I don't think there was too much work involved in upping 'Display Postscript' to 'Display PDF'. And I think the dropping of the license fee from Postscript to PDF was probably the driving factor, moreso than the technical issues mentioned.
As for the ability to reference previously drawn objects... that's what widgets do in 2d bitmapped space. Its been possible for some time (see GLUT) to implement a 2d interface, with widgets, etc, in OpenGL, which can be scaled just as easily, retains as much information, etc.
Really, I would have to say that for the Linux community, SGI's actions regarding OpenGL, and they're apparent attempts to pave the way for opensource OpenGL accelerated hardware drivers is more important. When we have hardware accelerated OpenGL drivers for most of the major cards, it would be relatively simple to create an OpenGL based WM, or OpenGL based apps, and get all the same abilities. The only realm that I know of where Postscript/PDF has a strong advantage over OpenGL is fonts, and considering the recent work done in integrating FreeType with OpenGL, I consider that an advantage likely to be shortlived.
As for the transparency effects, those are easy in OpenGL... but also here today in imlib2 and gdbpixbuf (actually, libart, I think).
When the Gnome guy said it was trivial, he was referring to AQUA, NOT QUARTZ!!! Aqua, i.e. dynamically scaling bitmaps through arbitrary transformations, and using transparency and truecolor widgets, is all possible today with libart and gdk-pixbuf, and possible tomorrow with hardware accelerated OpenGL.
I don't think Linux has much to worry about here.
(Speed and efficiency issues aside, which are currently being remedied. Most of the GNOME complaints earlier in the discussion can be laid at the feet of stacking libart, et. al., on top of imlib, which was designed and optimized for a different purpose. And OpenGL merely awaits hardware acceleration with capable drivers.)
(ot) Can /. team agree on what to post and when? (Score:3)
What Slashdot readers should be given the ability to do is read through article submissions. Slashdot fancies itself as a censorship-free society, but truly, whenever you have people selecting which information the group can read, you are making decisions that censor. I don't have an Ars box and I don't read Ars regularly, so I would never have known about this article, yet it's the most informative, well-written piece I've seen linked in a long time. Thank God that someone reposted it. When that Slashdot squirrel denied the post, he basically made a decision that said, "This isn't worthy of the general populace," which, to me, is an opinion-based decision that censored.
I'm sure Taco or Hemos could very quickly implement the functionality to read through the posts. They could even put it in a Slashbox. Basically, if the article has a URL, pull it out and check it (perl has this functionality via modules; it's very easy). If there's not already an article with that URL in the system, add it to a database, if that URL isn't already in the database. If the article is posted, mark it as posted. If it is rejected, mark it as rejected and make it available for viewing. If it is offensive for some reason (one of the reasons that the FAQ gives for not having such a list), just delete it altogether. The squirrels go through each and every article anyway, so this management must take place anyway. All they have to do is mark it. The database can store simply the subject and the URL (if they don't have a URL, tough luck), and then present a list of link titles that take people to various articles. Don't give people commenting rights on these posts. Only store rejects for, say, two weeks, to keep the database file small. I'm sure some of this functionality is already in (they have to store information for the squirrels in some manner), so implementing this on top shouldn't be too hard.
Maybe I should mail this to Rob. Eh, maybe not. He's a busy man.
Off-topic, or maybe not (Score:5)
Date: Sun, 30 Jan 2000 18:13:17 -0600
From: johnc@idsoftware.com (John Carmack)
Subject: debugging
.
.
.
BTW, the current status is that everything is compiling (all client apps work with remote X servers) with minimal porting work, and I have a building X server for OS-X using interceptor. The X server doesn't actually WORK yet (hence the debugging), but I don't think it will be too
long.
After I get the OS-X server version tested, I will just need to rewrite one file to hook into the darwin device drivers.
I'll leave it up to someone else to get gnome or kde ported.
John Carmack
Vector-based graphics: when are they coming? (Score:3)
Is anyone working on a vector-based system for Linux?
Does KDE like the sound of vector-based icons?
Does GNOME like the idea of having vector-based commands for their canvas project?
Do those other GUI systems, Berlin and GGI have any thoughts on a vector-based system?
Although I don't like many of the things that Mac OS X has, I do think ArsTechnica is right that a vector-based GUI is a step above what we have currently.
What does everyone here think?